Pinnacle West Capital Corporation (NYSE: PNW) — A Boomtown Franchise That Cannot Earn Its Allowance

Independent equity research Date: 25 July 2026 Price at analysis: $106.06 (close, 24 July 2026) · Market cap: ~$12.9bn · Dividend yield: ~3.4% Sector: Utilities · Regulated Electric (single-state, Arizona)

With the single, clearly-labelled exception of the Claude's Take block immediately below, the analysis in sections 1–15 carries no buy/sell recommendation and no price target. This article is general information, not investment advice.

⚡ Claude’s Take

This block is the author’s own subjective opinion. It is general information, not investment advice, and it is the single place in this article where a position is taken. Everything below it (sections 1–15) is position-free and target-free.

Verdict: HOLD — and AVOID adding here at ~$106. Accumulate only in the high-$80s to mid-$90s. Framing: a genuinely great service territory attached to a structurally impaired earnings machine, priced at its richest-ever multiple four months before a binary regulatory vote. Tag: “The best growth in American utilities, earned by the wrong shareholders.”

Pinnacle West owns something most utilities would pay dearly for: metro Phoenix. Weather-normalised retail sales have grown 4–7% for nine consecutive quarters, residential customers are compounding at ~2.4% (roughly double the national average), and the company has 4.5 GW of committed large load against an 8,648 MW system peak — a more than 50% increase in peak demand from customers already in hand, anchored by TSMC’s fab complex. That is not a story; it is in the delivered numbers. And it comes with two underrated structural gifts: Arizona’s H.B. 2201 wildfire safe harbour, which is close to the mirror image of California’s inverse-condemnation regime that dominates analyses of PG&E and Edison International, and a FERC-regulated transmission book earning 10.75% on a formula rate with annual true-up — the fastest-growing and by far the highest-quality dollar the company spends.

And it has converted almost none of it. Over 2021–2025 PNW put $9.9bn of gross capex to work, grew net plant 29.4% and operating income 32.6%, and delivered net income to common that was lower in 2025 ($616.5m) than in 2021 ($618.7m). Diluted EPS fell from $5.47 to $5.05 while the share count rose 7.8%; return on incremental invested capital was approximately zero. APS earned a 7.78% GAAP ROE in FY2025 against a 9.55% authorisation — a ~177bp, ~$152m, ~$1.24-per-share shortfall — and management’s own stated goal is to be earning within 50bp of authorised only by 2029, conceding “structural lag that will continue to exist” thereafter. The causes are structural, not cyclical: a historical test year with a 16–28 month filing-to-rates lag, Arizona’s fair-value rate-base construct that produced an effective return of just 4.39% in the last case, ~$1.9bn of holdco debt that earns no regulated return, and a load mix in which management concedes a point of residential growth is worth “north of $25 million” of margin while a point of data-centre load is worth “$5 million to $10 million.” The headline growth is real; it is simply 3–5x less profitable than it looks.

Against that, the market pays 22.8x guided FY2026 EPS of $4.55–$4.75 — a guided-down year — at 1.86x book and a 98.5th-percentile composite valuation versus the stock’s own decade. Essentially 100% of the +21.8% year-to-date move is multiple expansion, and the factor work says the momentum is borrowed rather than earned: PNW’s stock-specific Momentum loading decays from +0.241 to +0.038 once the sector is stripped out, while the “Regulated Utility Giants” basket it rides sits at a +2.84 z-score over the trailing year. The offsetting bull case is real and I do not dismiss it — a Formula Rate Adjustment Mechanism would genuinely cure the lag, and both APS and ACC Staff have now put FRAM parameters on the record. But that mechanism is sub judice: the Arizona Court of Appeals remanded the ACC’s formula-rate policy statement in November 2025 and review is pending at the state Supreme Court. Meanwhile the ROO lands in November 2026 and the vote in December, before a directly elected commission whose chairman was defeated in his own party’s primary on 21 July 2026 — four days before this report — by a challenger running against utility rate increases, while the state Attorney General has intervened in opposition to the increase and RUCO is at 9.0–9.2% against a current 9.55%. The last time Arizona voted a rate case in a hostile political climate, in November 2021, it handed APS a negative revenue outcome and an 8.70% ROE, and the stock fell 8.07% in a single session. When the favourable 2024 order arrived, the stock fell ~4.7% over the following week. This regulator has been priced as strictly one-sided risk, and today’s multiple leaves no room for that asymmetry to reassert itself.

Conviction: medium. Flips bullish if the December 2026 order grants a workable FRAM with an ROE at or above 9.55% and a revenue increase near $550–600m, and the Arizona Supreme Court affirms the formula-rate policy — that combination makes 5–7% EPS growth real for the first time and would justify re-underwriting at a higher multiple. Flips bearish if the order cuts the ROE below 9.55% or denies the FRAM: that is 2021 repeated, but with ~$1.9bn of holdco debt, a thinner FFO/debt cushion, five points of covenant headroom, and a multiple 40% higher than it was then.

📈 Stock Price Action — Five-Year Event Map

Pinnacle West has round-tripped from a regulatory disaster to an all-time high. The stock bottomed at an adjusted close of $50.84 on 12 October 2022 and set an all-time-high close of $109.37 on 2 July 2026, closing $106.06 on 24 July 2026 — 3.03% off the all-time high (reconciling exactly to FactorsToday’s independent rs_peak of −3.03%), inside a 52-week range of $83.42–$109.37, +55.6% over five years and +21.8% year-to-date. Realised volatility has compressed into the rally and the deepest 12-month drawdown was just −8.4%.

Cycle narrative. (1) The worst session in five years was 7 October 2021, −8.07% close-to-close ($60.14 → $55.29), following the ACC’s vote on the 2019 rate case — a rare negative revenue outcome and the lowest authorised ROE of any major US utility at the time. (2) The stock recovered roughly a quarter into April 2022 as APS appealed and defensives outperformed a falling market. (3) That reversed into the five-year low as the 2022 hiking cycle repriced every bond proxy. (4)–(5) The 2023 round-trip was almost purely rates: +48% into July 2023, then −20% as the 10-year approached 5%. (6) The dominant up-leg began at the February 2024 low — and, critically, before the favourable order rather than because of it. Decision No. 79293, restoring a 9.55% ROE, was approved 22 February 2024 and the stock fell ~4.7% over the following week ($64.08 on 22 Feb to $61.06 on 28 Feb); the +47% that followed was driven by the load-growth narrative, with +3.97% on the 1 Aug 2024 print and +3.91% on the 6 Nov 2024 print. (7) The stock then went nowhere for ten months and did not react to the June 2025 rate filing. (8) The final leg to the all-time high runs entirely on load growth and the capital plan.

The asymmetry in events (1) and (6) is the single most instructive fact in this table and is carried into the variant-perception section: the punitive 2021 order cost 8% in a day; the favourable 2024 order was worth nothing. Arizona regulatory risk has been priced as strictly one-directional.

1. Executive Summary

Pinnacle West Capital Corporation is a Phoenix-based holding company whose earnings come essentially entirely from Arizona Public Service, the state’s largest electric utility: ~1.43m retail customers across 11 of Arizona’s 15 counties, ~$30.0bn of consolidated assets, and 95% of revenue from ACC-tariffed retail sales. It is a pure-play, single-state, vertically integrated regulated electric utility with no unregulated growth engine and no geographic diversification of regulatory risk.

The demand backdrop is genuinely top-tier. Weather-normalised retail sales grew 5.0% in FY2025 and 7.4% in Q1-2026 (C&I +14.6%), the ninth consecutive quarter at or above the original 4–6% guidance range. Residential customers grew 2.4% in 2025, roughly double the national average. Management guides long-term weather-normalised sales growth of 5–7% through 2030 and discloses 4.5 GW of committed large load against an 8,648 MW 2025 system peak, plus a ~20 GW uncommitted queue. Against the twenty-four regulated utilities used as a comparison set here, PNW’s delivered load growth ranks at or near the top.

The conversion of that demand into shareholder earnings has failed. Over FY2021–FY2025 PNW deployed $9.9bn of gross capex, grew net utility plant 29.4% and operating income 32.6%, and produced net income to common of $616.5m in 2025 versus $618.7m in 2021. Diluted EPS fell 7.7% (from $5.47 to $5.05) as average diluted shares rose 7.8%. The bridge is unambiguous: the +$262.3m operating-income gain was more than consumed by an 80.9% rise in interest expense (+$188.7m) and the permanent loss of ~$100m of pension/OPEB non-service credits (prior-service credits fully amortised 31 January 2025). Return on incremental invested capital was approximately zero. FY2026 EPS is guided to $4.55–$4.75, a further ~8% decline from the FY2025 actual.

The cause is regulatory structure, not operational failure. APS earned a 7.78% GAAP ROE in FY2025 against a 9.55% authorisation — a ~177bp / ~$152m / ~$1.24-per-share shortfall. Arizona uses a historical test year with a 16–28 month filing-to-rates lag; layers a fair-value rate base on top of original cost while granting only a 0.25% return on the increment (producing an effective fair-value rate of return of 4.39% in the last case); and the holding company carries ~$1.9bn of debt whose ~$90m of annual interest earns no regulated return and is structurally unrecoverable — roughly 99bp of consolidated ROE.

Load growth is also lower-quality than the headline. The CFO has quantified it: a point of residential growth is worth “north of $25 million” of margin; a point of extra-high-load-factor (data-centre) growth is worth “$5 million to $10 million.” Since 4–6 points of the guided 5–7% is large C&I, the headline overstates the earnings contribution by a wide margin — which resolves the central paradox of nine strong sales quarters alongside a negative five-year EPS CAGR.

Everything now turns on one dated decision. The 2025 Rate Case (docket E-01345A-25-0105, filed 13 June 2025) asks for a $579.5m / 13.99% net increase (rebutted to $611.3m / 14.75%), a 10.70% ROE, a 1% fair-value return, and a Formula Rate Adjustment Mechanism that would convert a lumpy historical-test-year regime into an annual true-up. The hearing opened 18 May 2026 and ran to early July; the ALJ recommended order is expected November 2026 and the Commission vote in December. The positions on the record: ACC Staff at $525.2m and 9.55–9.80%, RUCO at $200.2–278.1m and 9.00–9.20%, and the state Attorney General intervening in opposition to the increase. (A widely-repeated claim that the Attorney General advocated a 6% ROE could not be corroborated to a primary source and is not relied on here.) The FRAM’s legal foundation is itself contested — the Court of Appeals remanded the ACC’s formula-rate policy statement in November 2025 and review is pending at the Arizona Supreme Court.

And this regulator is elected. Arizona is one of ~14 states electing its utility commissioners. The five-member ACC is currently 5-0 Republican, and on 21 July 2026 — four days before this report — sitting Chairman Nick Myers was defeated in his own party’s primary by a challenger running against utility rate increases. RRA lowered Arizona to Below Average, its lowest category, after the 2021 rate case and has flagged commissioner turnover as the specific reason to watch the state.

Two genuine, under-appreciated positives cut the other way. Arizona’s H.B. 2201 (signed May 2025, effective September 2025) limits utility wildfire liability where the utility complies with an approved mitigation plan absent clear and convincing evidence of willful or reckless misconduct, and permits securitisation — close to the inverse of the California regime that dominates analyses of PG&E and Edison International. And the FERC transmission book earns 10.75% on a formula rate with contemporaneous annual true-up, is free of Arizona lag, and is the fastest-growing capex bucket (+56% from 2026 to 2028), contributing $0.16 of the $0.31 Q1-2026 EPS improvement.

Valuation is the binding problem. At $106.06 the stock trades at 21.0x FY2025 actual and 22.8x the FY2026 guidance midpoint, at 1.86x book, and at a 98.5th-percentile composite valuation versus its own decade (P/E 97.6th, P/B 99.3rd, P/S 98.6th). Its apparent 14% trailing-P/E discount to the peer median largely disappears on forward guided earnings, because peer EPS is rising and PNW’s is falling. Effectively all of the +21.8% year-to-date move is multiple expansion, and the factor evidence indicates the momentum is borrowed from a crowded sector basket (Regulated Utility Giants at a +2.84 trailing-year z-score) rather than earned at the company level. The market is simultaneously underwriting that the AI load is real, that the ACC will materially close a decade-old earning gap, and that the formula rate survives judicial review. Those are three independent bets, and the last two must clear an elected commission in December.

2. Business Overview

2.1 Structure

Pinnacle West Capital Corporation (NYSE: PNW) is a Phoenix, Arizona holding company with ~$30.0bn of consolidated assets at 31 December 2025 and one reportable segment: regulated electricity. The 10-K states plainly that “we derive essentially all of our revenues and earnings from our principal subsidiary, APS.”

Arizona Public Service Company (APS), founded 1886, serves ~1.43m retail customers in 11 of Arizona’s 15 counties. It employs 6,610 people, including ~2,400 at jointly-owned generating facilities APS manages; ~1,120 are IBEW-represented under a collective bargaining agreement with an amendable date of 1 April 2026, and IBEW Local 387 has given notice of a request to bargain — a live 2026 cost item.

The only other subsidiaries are immaterial: El Dorado Investment Company (minority and debt investments in energy-related and Arizona community ventures, including a ~$21m equity-method stake in SAI Advanced Power Solutions, a switchgear manufacturer serving data centres) and PNW Power, which holds the assets of the former Bright Canyon Energy, sold in 2024 — a 50/50 TransCanyon joint venture with Berkshire Hathaway Energy’s BHE U.S. Transmission pursuing the 214-mile, 500-kV Cross-Tie line between Utah and Nevada (BLM Record of Decision 18 December 2025, non-appealable from late January 2026), plus 9.9% of the 242 MW Clear Creek wind farm (equity investment fully impaired in Q4 2022) and 5.1% of the 250 MW Nobles 2 wind farm.

The non-utility stub is immaterial to earnings and its record is poor: Bright Canyon divested, Clear Creek written to zero. For analytical purposes PNW is a 100% pure-play single-state regulated electric utility. There is no diversification and no hedge against a single regulator — a point that matters more here than in most utility files, because that regulator is elected.

2.2 Revenue and customer mix

FY2025 total operating revenues were $5,339.9m:

Retail electric revenue was 95% of FY2025 revenue (three-year average ~94%). FY2025 retail volumes were 34,198 GWh (residential 14,922; business 19,276) against total electric sales of 38,377 GWh. Derived average realised prices: 14.87¢/kWh retail blended, 17.03¢/kWh residential, 13.19¢/kWh business.

The customer economics are worth stating explicitly because they drive the growth section. Residential accounts for 89.8% of retail customers (1,287,097 of 1,433,293) but only 43.6% of retail kWh and 50.0% of retail revenue. A business customer averages ~131,364 kWh/yr (~$17.3k); a residential customer ~11,493 kWh/yr (~$1,957, about $163/month).

Customer concentration is currently negligible — “during 2025, no single purchaser or user of energy accounted for more than 1.9% of our electric revenues.” But that is a backward-looking statistic that the company’s own plan destroys: with 4.5 GW of committed large load against an 8,648 MW peak, and 4–6 of the guided 5–7 points of sales growth coming from large C&I, essentially all incremental volume arrives from a small number of data-centre and semiconductor counterparties. The 1.9% disclosure will not survive the plan, and the risk section treats the resulting concentration as a rising risk.

Credit quality of the receivable book is good: FY2025 bad debt expense was $28.6m on $5.34bn of revenue, or 0.54%.

2.3 The generation fleet

APS owns or leases 6,257 MW of regulated generating capacity against a FY2025 one-hour system peak of 8,648 MW (7 August 2025, up from 8,210 MW on 4 August 2024). Owned capacity is therefore only 72.4% of peak — APS is structurally short capacity and covers the gap with tolling PPAs and market purchases. This matters: it means the load-growth story requires a build, and the build is happening into an inflating turbine market.

Nuclear — Palo Verde Generating Station. The three-unit plant ~50 miles west of Phoenix is the largest power producer in the United States by annual generation — a precision worth keeping, because at ~3.9 GW of capacity it was overtaken on a nameplate basis by Vogtle (~4.5 GW) when Units 3 and 4 entered service in 2023–24. PNW’s own language is careful on this point and this memo follows it. APS operates it and holds 29.1% of Units 1 and 3 and ~23.9% of Unit 2 plus a ~5.2% leasehold (29.1% combined), for a 1,146 MW entitlement — the largest of the seven participants (others: Salt River Project, Southern California Edison, El Paso Electric, PNM, SCPPA, LADWP). NRC licences run to June 2045 (Unit 1), April 2046 (Unit 2) and November 2047 (Unit 3), and APS has filed its intent to seek renewal into the mid-2060s. FY2025 owned nuclear output was 9,193 GWh at a 92% capacity factor (95% in 2024), and management reported a 100% summer capacity factor in 2025. In June 2025 APS agreed to buy out two of the three Unit 2 sale-leaseback interests (~7% of Unit 2, ~94 MW) for ~$199m, closing September 2025 and leaving one lease (~5.2% of Unit 2) expiring 2033. Palo Verde decommissioning is recovered through a non-bypassable system benefits charge paid by all retail customers on the APS system.

Palo Verde is the single best asset in the business: 22.7% of delivered energy at nuclear fuel cost, firm and carbon-free, with licences running two decades beyond the current planning horizon — precisely the profile large-load customers now pay up for.

Coal — in retreat. Four Corners Units 4 and 5 in north-west New Mexico: APS operates and owns 63% for a 970 MW entitlement (co-owners SRP 10%, PNM 13%, Tucson Electric Power 7%, NTEC 7%), with coal supplied by NTEC under an agreement to 2031 and the Navajo Nation facility lease extended to 2041. A seasonal-operation option agreed in June 2021 and retained in July 2024 has not been exercised “due to market conditions.” FY2025 output 6,021 GWh at a 65% capacity factor. Cholla: coal-burning ceased March 2025 and Units 1 and 3 were formally retired 30 April 2025 with ~$81m of remaining net book value; recovery of remediation, CCR closure and remaining plant costs is being sought in the 2025 Rate Case after a standalone deferral request was withdrawn on 8 July 2025. Cholla Unit 2 (closed 2015) carries a $23.6m regulatory asset amortising through 2033. Navajo Plant (14% APS interest, 315 MW, SRP-operated) ceased operations in November 2019, leaving $23.8m of book value plus a $2.5m coal reclamation regulatory asset, recoverable except for 15% of annual amortisation.

Gas and oil — 3,722 MW across Redhawk combined-cycle (1,140 MW), West Phoenix CC (874 MW), Ocotillo CT (630 MW), Sundance CT (520 MW), Saguaro CT (189 MW), Yucca gas and oil CTs (237 MW), West Phoenix CT (114 MW) and Douglas oil CT (18 MW). FY2025 output 10,197 GWh at a 32% capacity factor. Two CTs (~90 MW) entered service at Sundance in 2025; eight CTs (~397 MW) at Redhawk are under construction for 2028. APS plans to add up to 2,000 MW of flexible gas generation (the two-phase Desert Sun project at Gila Bend, Phase 1 late-2030) and is an anchor shipper on the Transwestern Desert Southwest pipeline (upsized from 42" to 48", 1.5 Bcf/d, in service 2029/30).

Renewables and storage — modest and shrinking as a share of plan. 419 MW of owned solar in operation (Ironwood 168 MW, Agave 150 MW, Red Rock 44 MW, Foothills 38 MW, Gila Bend 36 MW, plus 41 MW distributed) with 168 MW under development, and 201 MW of owned battery storage with 150 MW under development. FY2025 owned renewable output 946 GWh at a 29% solar capacity factor. A large renewable PPA book includes Solana 250 MW CSP, Sunstreams 3 and 4 (215/300 MW), Harquahala Sun 2 (300 MW), Serrano (170 MW) and 2026–27 additions including Hashknife 1 and 2 (275/200 MW), Catclaw (225 MW), CO Bar Solar C (206 MW) and Papago (150 MW). Non-renewable tolling agreements of 600 MW, 600 MW and 525 MW run to 2032/2034/2038.

Energy mix, honestly stated. FY2025 total energy sources were 40,423 GWh: owned generation 26,358 GWh (65.2%) and purchased power 14,065 GWh (34.8%). By source: gas/oil/other 25.2%, nuclear 22.7%, purchased renewables 16.4%, purchased conventional 15.3%, coal 14.9%, resales 3.1%, owned renewables 2.3%. Carbon-free delivered energy is therefore ~41.5%, not the “approximately 58%” management cites — because the company’s definition folds in demand-side management, which is avoided consumption rather than supplied energy. Treat the 58% as a marketing construct. Note also that in August 2025 APS downgraded its target from “zero-carbon by 2050” to “carbon-neutral by 2050,” permitting offsets.

2.4 How the money is actually made

APS earns a regulated return equal to rate base × authorised equity layer × allowed ROE, plus a debt return, plus — uniquely in Arizona — a small statutory return on the increment of fair value rate base above original cost.

Two very different regulated businesses sit inside APS:

This split is under-appreciated and recurs throughout the memo. The FERC book earns 120bp more, suffers no Arizona lag, and is guided to grow from ~$2.52bn (2024) to ~$4.0bn (2028), a 59% increase versus 28% for the ACC book. It is the highest-quality dollar PNW spends.

Adjustor mechanisms recover certain costs between rate cases. The PSA (Power Supply Adjustor) passes through retail fuel and purchased-power variances, resets each 1 February, and is capped at ±$0.006/kWh of annual change (raised from $0.004 in Decision 79293); the deferred fuel regulatory asset has unwound from $463.2m (2023) to $287.6m (2024) to $149.1m (2025), and the PSA rate stepped from $0.011977 to $0.013977 to $0.016977/kWh. The LFCR recovers fixed costs lost to energy efficiency and rooftop solar at 2.56¢/kWh residential, capped at 1% of retail revenues annually and — importantly — contingent on an annual earnings test that reduces recovery if APS over-earns; the approved amount was $60.1m for 2025, implying roughly 2,348 GWh (~6.9% of retail volume) of annual lost sales. The TCA passes FERC formula-rate transmission changes into retail rates each 1 June, and delivered a step-change in 2025: the FERC annual transmission revenue requirement rose ~$119.0m (versus $27.4m in 2024), of which $88.3m flowed to retail. The RES surcharge (2026 budget ~$110.1m) and DSMAC fund renewable and efficiency programmes — the ACC cut the DSM budget to $40m on 3 December 2025 and ordered refunds of uncommitted funds in April 2025 (APS refunded $44.2m RES and $7.6m DSMAC). The TEAM is currently set to zero. The CRS (Court Resolution Surcharge) at $0.00175/kWh recovers $59.6m of revenue lost between December 2021 and June 2023 following an Arizona Court of Appeals reversal ($43.2m collected at 31 December 2025).

What the adjustors do not cover is the heart of the problem: base O&M inflation, wildfire mitigation O&M and insurance (the ACC denied a wildfire deferral on 18 June 2025 and pushed it into the rate case), depreciation on new plant, financing costs, and the return on incremental rate base between cases. Those are precisely the costs that are inflating, and they are recovered only when a case is decided.

Verdict (the business overview). The revenue base is exceptionally high quality: an essential, non-discretionary, non-storable product sold into a 100%-share territory with two decades of ~2%+ customer growth in one of America’s fastest-growing metros, negligible bad debt, and a genuine large-load demand shock anchored by TSMC. The cost structure is more mixed than the fleet table suggests — only 72.4% of peak is owned, 34.8% of delivered energy is purchased, and the company is exiting coal into a gas build at a time of severe turbine cost inflation. The best asset by some distance is the operated 29.1% Palo Verde interest. Quality of the revenue base: high. Quality of the earnings stream that revenue base produces: materially worse — and that is a regulatory question, answered in the competitive-position section.

3. Industry Dynamics

3.1 The profit-pool mechanic — capex is the product

A US vertically integrated regulated electric utility earns, in identity form: Allowed Net Income = Rate Base × Equity Ratio × Allowed ROE. Everything else — revenue, fuel, purchased power, DSM, renewable-standard spending — either flows through as a pass-through or is a cost the regulator allows into the revenue requirement. APS’s 2025 Rate Case makes the identity explicit: a $12.5bn original-cost rate base on a 2024 test year, a proposed 52.35% equity / 47.65% debt structure at a 4.26% embedded debt cost, a requested 10.70% ROE, a 7.63% WACC, plus a 1% return on the fair-value increment.

Because the allowed return is a fixed percentage of a regulated asset base, the only structural lever on earnings is the size of that base. There is no volume leverage (fuel is a pass-through via the PSA), no price leverage (rates are set by the regulator), and no durable margin leverage (O&M savings are largely handed back at the next case). A regulated utility does not sell electricity for profit; it sells capital deployment to a regulator, and electricity is the delivery mechanism. APS plans $2,600m / $2,650m / $2,700m of capex in 2026/2027/2028 — $7.95bn over three years — against FY2025 operating cash flow that does not cover capex plus the dividend. That external-funding signature is the definition of the model, not a defect of this company.

3.2 The Marathon capital-cycle inversion — and where the risk actually lives

Marathon’s capital cycle holds that high returns attract capital, capital erodes returns, the discipline is competitive entry and the punishment is falling ROIC. In a regulated utility this mechanism is structurally disabled in both directions, and understanding that is essential to reading PNW correctly.

The upside is capped. The regulator sets the allowed ROE at roughly the cost of equity. There is no scenario in which APS earns 25% on incremental capital; the best case is that it earns its authorised return. Consolidated ROIC for a rate-base utility runs ~5% — below any sensible WACC — and this is normal, not evidence of value destruction, because the model deliberately finances a large asset base with cheap debt and earns the allowed return on a thin equity layer. Any analysis that condemns PNW simply because ROIC is 4.6% has misunderstood the industry.

The downside discipline is also disabled. APS is a legal monopoly; no competitor can enter. Capital deployed into rate base is not competed away, it is rewarded with an incremental earnings stream. In Greenwald’s taxonomy this is not a supply advantage, not customer captivity, and not economies of scale — it is government-conferred protection, the most absolute barrier to entry in the framework and simultaneously the least valuable, because the same authority that erects the barrier sets the return behind it.

Therefore the capital-cycle risk is relocated, not removed. What mean-reverts a regulated utility’s returns is not competitive entry but regulatory and customer-affordability backlash. The binding constraint on rate-base growth is not the market’s ability to absorb supply; it is the customer bill’s ability to absorb the revenue requirement, mediated through a political body. When rate base compounds faster than bills can politically absorb, the regulator does not stop the capex — it cuts the allowed ROE, disallows the capital, extends the lag, or all three.

The Marathon warning signs are all present in US utilities in 2026 — rising capex and depreciation, sector-wide capital influx, sell-side enthusiasm, index-weight expansion. But the read-across is not “returns will be competed down.” It is: the regulatory ask is growing faster than the political system’s tolerance, and the adjustment arrives through the rate case. APS is asking for a 14.75% net revenue increase in a state where the regulator stands for election. That is the capital cycle in its regulated form, and Arizona is the jurisdiction where it bites hardest.

3.3 Regulatory lag is the structural tax

Arizona uses a historical test year. The 2025 Rate Case rests on a FY2024 test year, was filed 13 June 2025, went to hearing 18 May 2026, and will not produce rates until early 2027 — roughly a 30-month gap between test-year midpoint and rate effectiveness, during which APS spends $2.6bn a year while earning a return on a rate base that is two and a half years stale. The 2022 case took ~16.5 months from filing (28 October 2022) to rates (8 March 2024); the 2019 case took ~25 months.

Management is explicit that this is the binding constraint. CFO Andrew Cooper, Q4-2025 call:

“It really is almost entirely about the timing consistency of cost recovery. We’ve got a pretty good view of our committed demand, and it’s robust. But due to the substantial regulatory lag in the jurisdiction, the ability to consistently on a linear basis translate that top line growth and the bottom line growth is challenged by the lumpy nature of our rate case process.”

Quantified: FY2025 capex of $2,624.6m against a 2024 test-year rate base of $12.5bn is ~21% of rate base added in a single year, while rates in effect through mid-2026 are built on the 2021 test-year rate base of $10.5bn. Roughly $2.0bn of rate base is not in rates as of the 2024 test year alone; at a 52.35% equity layer and 9.55% ROE that is ~$100m after tax, or ~$0.82 per share of earnings the company is entitled to but is not collecting.

3.4 Arizona’s fair-value rate base — the quirk that peer comparisons miss

Arizona layers a fair value rate base on top of original cost, then grants a very small return on the increment: 0.25% in the 2022 case (1% requested in 2025). The consequence is that even when the headline authorised ROE looks competitive, the effective fair value rate of return was only 4.39% in the last case.

This matters enormously for benchmarking. RRA/S&P data put the median authorised ROE in US electric rate cases at 9.75% (Q1-2025) and 9.70% (FY2024), with an average award of 9.66% in 2025; recent high-water marks include Florida Power & Light at 10.95%. On headline, APS’s 9.55% sits only ~10–20bp below the national median and looks unremarkable — better than Ameren Illinois (8.72%) or ComEd (8.905%). But no peer jurisdiction carries the fair-value drag. Any comparison that reads “9.55%, roughly normal” and concludes Arizona is an average jurisdiction is wrong, and it is a mistake the sell-side framing routinely makes.

3.5 The elected Arizona Corporation Commission

Arizona is one of only ~14 states whose utility regulators stand for statewide partisan election rather than gubernatorial appointment. Following a 2000 ballot measure, the ACC has five members serving staggered four-year terms with a two-consecutive-term limit. As of July 2026 the Commission is 5-0 Republican: Chairman Nick Myers (elected 2022, chair from January 2026), Vice Chair Rachel Walden (2024), Kevin Thompson (2022, chair in 2025), Lea Márquez Peterson (appointed 2019, re-elected 2024, term-limited) and Rene Lopez (2024). Republicans swept all three seats in November 2024.

On 21 July 2026 — four days before this report — sitting Chairman Nick Myers was defeated in the Republican primary by state legislator Ralph Heap: Thompson 35.82%, Heap 32.98%, Myers 31.19% — a ~1.8-point margin, roughly 15,000 votes, with 100% of precincts reporting in all fifteen counties. Myers conceded, noting he would “finish my term” over “the next 5 months.” Heap and fellow challenger David Marshall were backed by the Legislature’s Freedom Caucus and campaigned against utility rate increases. Thompson and Heap face Democrats Jonathon Hill and Clara Pratte in the November 2026 general election. Separately, Arizona Capitol Times reported on 15 July 2026 that the incumbents had accepted “prohibited donations.”

This is the most important industry fact in the whole analysis and it is absent from the bull model. The APS rate case ALJ recommended opinion is expected November 2026 with a Commission vote in December — taken by a lame-duck chairman who was just repudiated in his own party’s primary by a candidate running specifically against rate increases, in the same season the general election is decided. There is no plausible reading in which a commissioner deciding a ~14% residential rate increase during an election cycle is insulated from the electorate. This is the structural difference between Arizona and virtually every appointed jurisdiction in the peer set (Michigan’s MPSC, Wisconsin’s PSCW, Minnesota’s MPUC, Texas’s PUCT, Illinois’s ICC, Florida’s FPSC, California’s CPUC), and it is why the same demand backdrop is worth materially less here. Comparable published utility analyses almost never treat this variable; Southern Company, before Georgia’s elected PSC, is the rare exception.

Where Arizona ranks. RRA (S&P Global) lowered Arizona to Below Average — its lowest category — following the November 2021 rate case, characterising the 8.70% award as among the lowest ROEs it had encountered for a vertically integrated electric utility in thirty years; Arizona had been Average/3 in 2016. RRA has separately flagged Arizona among a small set of states to watch for near-term shifts in regulatory risk, citing commissioner turnover — which materialised on 21 July 2026. Arizona’s current (2025–26) sub-tier could not be confirmed to a primary source, since RRA’s state evaluations sit behind an S&P Global subscription; secondary indications of an upgrade within the Below Average band after the constructive February 2024 order are plausible but unverified, and this memo does not rely on them. What is not in doubt is the direction of travel from a low base: independent analysis put Arizona’s five-year average approved ROE below all but four states and its regulatory lag at roughly 590 days, longer than any US jurisdiction except the District of Columbia.

An important fairness correction: “elected” is not, by itself, the explanation, and this memo does not assert it as one. Georgia’s PSC is elected and authorises 10.5–11.5%; Louisiana’s is elected and authorises ~9.7% with a band. Election alone plainly does not produce bad outcomes. Arizona’s problem is a stack of four features, only the last of which is about ballots:

1. A historical test year with a 14.6–25.0 month filing-to-rates clock and no forward-test-year option.
2. A statutory power to reopen final orders. Arizona “has a statute that allows the ACC to reopen prior decisions and modify otherwise final orders” (PNW’s own risk factor). In most jurisdictions a final rate order is final.
3. An unresolved constitutional and jurisdictional overhang — the 2004 Court of Appeals line, the live formula-rate rulemaking challenge, and the 10-K’s own reference to the ACC’s “permissive regulatory authority” being called into question.
4. An elected, affordability-populist commission with a demonstrated instance, within the last five years, of imposing an explicit ROE penalty and a nine-figure prudence disallowance that an appellate court had to reverse.

And the volatility runs in both directions, which is the point. In March 2026 the Commission voted 5-0 to repeal Arizona’s Renewable Energy Standard (Decision No. 81677) — a decision squarely in APS’s favour. It has not taken effect. The Attorney General disapproved the Notice of Final Rulemaking under A.R.S. the relevant section-1044, arguing the Commission approved publication before completing the required economic impact statement; the ACC has authorised its general counsel to litigate, calling the move an attempt to “essentially veto” its ratemaking authority. PNW’s own Q1-2026 10-Q states flatly: “APS cannot predict the outcome of this matter.” The energy-efficiency repeal reached a final 4-0 vote on 8 July 2026 and was routed to the Secretary of State without a disclosed Attorney General step — apparently on the theory that it is plenary ratemaking and exempt from certification.

This matters far beyond the rules themselves, because it is the same fight as the formula rate. Three separate 2025–26 disputes — the renewable-standard repeal, the formula-rate policy challenge remanded in November 2025, and the procedural route chosen for the efficiency repeal — all turn on one question: how far the Commission’s exclusive Article 15 ratemaking authority extends against the Administrative Procedure Act’s procedural overlay. That is the identical the relevant section-1044 certification defect that invalidated seven of Arizona’s retail-competition rules in Phelps Dodge in 2004. Twenty-two years on, the same choke point is now the single largest source of uncertainty over whether APS gets its formula rate. An investor underwriting the FRAM is underwriting the ACC’s side of an unresolved separation-of-powers dispute with the state’s Attorney General.

It is the stack, not the ballot, that produces a 4.6% ROIC. And the ACC has done several unambiguously pro-utility things: the 2024 order granting $491.7m and 9.55%, the formula-rate policy itself, the Grid Access Charge (since vacated on appeal, below), annual maximum 10% cuts to the solar export rate, closing the distributed-generation docket, and voting to repeal the renewable-energy and energy-efficiency mandates (though the renewable repeal is currently blocked — below). The accurate word for this regulator is not “hostile” but “volatile and unpredictable” — which is precisely what APS’s own cost-of-capital witness identifies as most damaging: “Credit rating agencies regard consistency and predictability of regulation as one of the most important factors.”

3.6 The rate-case record, and the dark-money legacy

The 2021 order is the precedent that matters. On 2 November 2021 the ACC decided the 2019 Rate Case with a total annual revenue change of negative $4.8m — a $120m cut to base-rate revenue, offset by adjustor transfers — and an authorised ROE of 8.70% (including a 20bp penalty later reversed to 8.90% via the Court Resolution Surcharge). The ALJ had recommended a $111m decrease and a 9.16% ROE; the Commission went below its own ALJ on ROE. Against a then-national average of ~9.4%, 8.70% was among the lowest awards in the country, and the order denied Four Corners SCR cost recovery.

The 20bp penalty had a specific, documented cause, and it traces directly to the 2017 rate case. It was imposed for customer-service failures — principally a faulty online rate-comparison tool that gave thousands of customers inaccurate information for roughly nine months (producing an Arizona Attorney General consent decree) and below-average call-centre performance found by ACC Staff. That in turn grew out of the backlash to the 2017 settlement: customer complaints that the actual bill impact exceeded the advertised 4.54% led the Commission to open a rate review in January 2019 and to hire an outside consultant, Overland Consulting, to audit APS’s books — a report the ACC subsequently withheld on privilege grounds. Separately, the September 2018 death of Stephanie Pullman, disconnected over roughly $51 of arrears on a 107°F day, produced the summer disconnection moratorium. The through-line from 2017 to the 2021 order is operational and reputational, not only political — which matters, because it means the risk is partly within management’s control rather than purely a function of who sits on the Commission.

The 2022 case partially repaired it. On 22 February 2024 the ACC approved a ~$491.7m base revenue increase, a 9.55% ROE, a 0.25% fair-value increment return, an effective fair-value rate of return of 4.39%, a WACC return on the net prepaid pension asset, and the SRB mechanism. Rates took effect 8 March 2024. A rehearing on the solar Grid Access Charge affirmed the GAC on 17 December 2024 but ordered a site-load cost-of-service allocation in the next case; SEIA, AriSEIA, Vote Solar, the Arizona Attorney General and two customers have appealed to the Court of Appeals, unresolved.

The dark-money legacy is the causal root of the 2021 order, not a historical curiosity. Pinnacle West anonymously routed ~$10.7m through third-party organisations in the 2014 ACC election cycle (including ~$3.2m supporting commissioners Doug Little and Tom Forese) and spent >$4m openly in 2016. Commissioner Bob Burns subpoenaed APS and Pinnacle West in August 2016; then-CEO Don Brandt refused disclosure on free-speech grounds; documents released in April 2019 confirmed the 2014 funding, including support for a sitting commissioner’s son. Brandt sold ~400,000 PNW shares for >$32m after the 2014 election. The company forswore ACC election spending under his successor.

The sequence is not coincidental: APS funds commissioners (2014–16) → the scandal breaks (2016–19) → the political cost is borne by a successor commission that must demonstrate independence → the 2019 case is decided in November 2021 with a negative revenue outcome and an ROE below the ALJ’s recommendation. In Marathon terms this is the regulated analogue of a capital-cycle bust: the company over-invested in the one input that determines its return — regulatory goodwill — the price of that input collapsed, and returns mean-reverted violently. Seven years on, the jurisdiction still prices APS’s political capital at a discount.

3.7 The Arizona demand backdrop

Metro Phoenix is among the strongest utility service territories in the United States. Maricopa County was ranked the top county for economic development in 2025 by Site Selection; Arizona ranks #1 nationally for semiconductor manufacturing per Business Facilities; Phoenix ranks #1 of 15 top growth markets for manufacturing per Newmark and #1 in Western industrial markets by sales and industrial development. APS residential customer growth has run 2.1–2.5% annually since 2021, roughly double the national average per the Itron Annual Energy Survey.

The anchor tenants are real and named: TSMC’s Phoenix fab complex and Amkor’s $7bn packaging plant (raised from $5bn). Critically, both are in APS territory — a distinction that matters because Arizona’s load is split with Salt River Project, a political subdivision of the State of Arizona that is not ACC-regulated and competes for the same growth. Intel’s Chandler/Ocotillo campus, by contrast, sits in SRP territory. Any analysis that credits APS with “Arizona’s semiconductor boom” wholesale is overstating it.

3.8 Structural sector factors

Wildfire — a genuine Arizona advantage. Arizona H.B. 2201, signed by Governor Katie Hobbs in May 2025 and effective September 2025, requires utilities to prepare biennial wildfire mitigation plans and limits liability for wildfire damages where the utility complies with its approved plan, absent clear and convincing evidence of willful, intentional or reckless misconduct. Companion legislation permits securitisation of wildfire costs. APS has submitted a Comprehensive Wildfire Mitigation Plan to the Arizona Department of Forestry and Fire Management; review was underway at the FY2025 10-K. This is close to the inverse of California’s inverse-condemnation regime — strict liability without fault — which is the dominant risk in analyses of PG&E and Edison International, where California was characterised as the most hostile large US utility jurisdiction on a risk-adjusted basis. For a utility in a hot, dry, wildland-urban-interface state, a compliance-based safe harbour is a material reduction in tail risk. Two caveats: the safe harbour is conditional on plan approval and compliance, and it is untested in Arizona courts. It reduces the tail; it does not remove it. Note also that this protection came from the legislature and governor, not from the elected ACC — and that the ACC separately denied a wildfire cost deferral on 18 June 2025.

Water and drought. Colorado River shortage declarations and Arizona’s 2023 groundwater ruling — which halted approvals of new subdivisions relying on groundwater in parts of metro Phoenix — are a genuine constraint on the residential growth leg. Notably, water was not mentioned once on any of the four earnings calls reviewed, which is itself a disclosure observation. Palo Verde is uniquely positioned here: it is the only large nuclear plant in the world cooled with reclaimed municipal wastewater rather than a natural water body.

Extreme heat. Phoenix summer peaks drive both the reliability obligation and acute political sensitivity around summer bills and disconnection policy — a sensitivity with direct history at APS following the 2018–19 heat-related disconnection controversy and the resulting rules.

Rooftop solar. Arizona is a top-tier solar state and distributed generation erodes the fixed-cost base; the LFCR mechanism recovers it after the fact, at 2.56¢/kWh residential, capped and subject to an earnings test. The Grid Access Charge has since been vacated: in mid-June 2026 the Arizona Court of Appeals struck it down on due-process grounds — it had never been disclosed in the 2022 rate-case public notice and APS had never requested it — and held the Commission improperly shifted the burden of proof onto challengers. Reconsideration motions filed by the ACC (24 June 2026) and APS (1 July 2026) were pending as of this report date and no mandate had issued, so the ruling is not yet final. APS proposes a higher, properly-noticed GAC in the 2025 rate case, which may cure the defect prospectively.

Supply chain. Gas turbine order books are effectively sold out into the 2030s, transformer lead times remain extended, and APS is committing to up to 2,000 MW of new gas plus an anchor-shipper pipeline commitment into that market.

Verdict (the industry section). The regulated electric industry is structurally attractive: a legally protected monopoly with an administratively set return, insulated from competitive mean-reversion, and currently enjoying the strongest demand backdrop in forty years. Arizona, however, is a structurally below-average jurisdiction, and the two verdicts must be kept separate. A historical test year with a 16–28 month lag, a fair-value construct that cut the effective return to 4.39%, an RRA Below Average rating, a demonstrated willingness to cut revenue and go below the ALJ’s own ROE recommendation, and — uniquely among the peer set — a directly elected commission now in the middle of an election that has already unseated its chairman on an anti-rate-increase platform. The demand is the best in the peer group; the jurisdiction is among the worst. That combination, not either fact alone, is the investment case.

4. Competitive Position

4.1 Naming the moat, precisely

APS holds a legally exclusive franchise to serve retail electric customers in its certificated territory. In the Greenwald taxonomy this is government-conferred protection — not a supply/cost advantage, not demand-side captivity, not economies of scale with captivity. It is the most absolute barrier to entry available: a competitor cannot enter at any price, because entry is unlawful.

Applying the tests honestly:

The market-share stability test is uninformative here. APS’s share of its franchise territory is definitionally ~100% and cannot vary. Greenwald’s >5pp-over-5-to-8-years threshold cannot discriminate. So the test must be reframed: the real question is not “can a competitor take share?” but “how durable is the regulatory compact, and how much of the economic rent behind the barrier does the shareholder actually keep?”

The ROIC test is where the moat fails to show up in financial outcomes. The standing test applied here is that a moat claim which cannot be tied to a financial outcome that would deteriorate without it is not a moat. Here the franchise is real and something would certainly deteriorate without it — but the franchise has not produced an adequate return for its owners:

APS earned 7.78% against a 9.55% authorisation in FY2025 — a 177bp shortfall worth ~$152m after tax, or ~$1.24 per diluted share. The monopoly is genuine; the economic rent behind it is captured by the regulator and the customer, not the shareholder. This is the central finding of the report and it is not a cyclical observation — the same gap has persisted, in varying size, for a decade.

The double-leverage paradox sharpens it. Consolidated ROE (8.93%) is higher than the operating subsidiary’s (7.78%) because Pinnacle West carries $1,915.4m of holdco debt and pushes it down into APS as equity: APS’s GAAP equity is $8,881.6m while PNW consolidated equity is only $7,046.5m, and the ~$1.84bn difference is the holdco debt. Parent net interest of ~$90m/yr (~$68m after tax) equals ~99bp of consolidated ROE that is structurally unrecoverable in rates. Double leverage currently flatters headline consolidated ROE and is precisely why PNW’s consolidated debt/capitalisation (~60%) far exceeds APS’s (~50%). Investors are buying a levered claim on a utility that under-earns its allowed return.

4.2 What genuinely is advantaged

Three things survive scrutiny and deserve credit:

The FERC transmission franchise. The transmission book earns 10.75% on a formula rate with an annual true-up — contemporaneous recovery, no Arizona lag, no historical test year. It is guided to grow from ~$2.52bn to ~$4.0bn of rate base by 2028 (+59%, versus +28% for the ACC book), transmission capex rises from $550m to $860m between 2026 and 2028, and management notes it has “doubled and then doubled again” over five years. It contributed $0.16 of the $0.31 Q1-2026 EPS improvement. This is the one part of PNW where the moat converts cleanly into an adequate, timely return, and it is systematically under-weighted in the bear case.

Palo Verde. An operated 29.1% interest (1,146 MW) in the largest power producer in the United States, at a 92% capacity factor, licensed to 2045–2047 with renewal filed into the mid-2060s, cooled with reclaimed wastewater in a drought-constrained state. In a market where large-load customers are paying premiums for firm carbon-free power, this is a scarce and appreciating asset — and one no competitor can replicate.

Affordability headroom — real, but thinner than the rate comparison suggests. APS rates rose an average of 2.67% annually over 2018–2025, against CPI of 3.62% and Phoenix CPI of 4.24%: real rates have fallen. Arizona residential rates run ~15.5¢/kWh, some 15–17% below the ~18.8¢ national average, and APS is around ~12.8¢. That is a genuine defence and it undercuts a naive “elected ACC therefore automatic denial” bear case.

But the comparison must be completed honestly, because it cuts both ways. Desert cooling drives ~1,075 kWh of monthly usage, so the average Arizona bill is roughly $164/month against a US average near $155. Arizona therefore has below-average rates and above-average bills — close to the worst combination for a utility asking for 14%, because the utility’s defence (“our rates are cheap”) is invisible to the customer while the customer’s complaint (“my bill is $300 in August”) is politically decisive. The political salience of that bill is not hypothetical: the 2018 death of Stephanie Pullman, disconnected over a $51 arrears at 107°F, produced an ACC disconnection moratorium that is now permanent (no disconnections 1 June–15 October, or above 95°F/below 32°F). Affordability headroom is a real asset here, but it is measured in bills, not rates, and on that measure the headroom is thin.

4.3 Threats to the franchise

Salt River Project — the most under-appreciated structural fact about APS. SRP is the Salt River Project Agricultural Improvement and Power District, a political subdivision of the State of Arizona that is exempt from ACC jurisdiction. Its board is elected biennially by landowners on an acreage-weighted basis — five acres equals five votes. It serves ~1.06m residential customers across Phoenix, Mesa, Chandler, Gilbert and Glendale — the geographic core of the metro — on FY2023 revenue of ~$4.02bn, and it is APS’s co-owner in both of its largest baseload assets, holding 20.2% of Palo Verde and 10% of Four Corners.

The comparison that matters is price, and it must be made on a consistent basis. On EIA Form 861 data for calendar 2024, APS’s residential rate was 16.45¢/kWh against SRP’s 13.46¢ — SRP is 18.2% cheaper, a gap that has widened from 14.5% in 2021. Across all customer classes APS was 14.53¢ against SRP’s 11.59¢, or 20.2% cheaper. And in the same window in which SRP’s board voted itself a net 2.4% increase (February 2025, effective November 2025), APS is litigating for a 14.69% day-one increase through a contested proceeding.

One nuance cuts against the simple reading and is worth stating, because it connects to the affordability point in the competitive-position section: SRP’s average residential customer uses about 17% more electricity (1,211 kWh/month against APS’s 1,034), so the bill gap is far narrower than the rate gap — $162.97 against $170.02 a month in 2024, just 4.1%. Normalised to equal usage, SRP’s bill would be roughly $139 against APS’s $170. The structural rate advantage is real and large; the lived bill difference is much smaller.

The source of the advantage is structural, not operational. SRP has no shareholders, no authorised ROE, no federal or state income tax, and Aa1/AA+ tax-exempt debt against APS’s Baa1/BBB+ taxable paper — its “equity” is $7.0bn of accumulated net revenues that costs nothing, and its only financial constraint is a bond rate covenant of roughly 1.0x coverage rather than a 10.70% requested return on a 52.35% equity layer plus a tax gross-up. It needs no commission’s permission to price: rates are set by its own elected board.

How the competition actually works matters, and it is not customer switching. The 1955 Territorial Agreement — still a listed material contract in PNW’s FY2025 10-K, alongside the Power Coordination Agreement and a 1998 Memorandum of Agreement — divides metro Phoenix along metes-and-bounds township lines, enforced by liquidated damages of 35% of the applicable charges for serving into the other’s territory, and explicitly survives municipal annexation. There is essentially no customer contestability. The competition runs through site selection: a new fab or data centre is served by whoever owns the ground it sits on, and the developer chooses the parcel. The revealed pattern is stark — the East Valley is SRP (Intel’s Ocotillo campus, Meta in Mesa, Google’s Redhawk campus, Apple, EdgeCore’s 450 MW), while the West Valley and north Phoenix are APS (TSMC, Microsoft, Vantage, QTS).

That is still a genuine ceiling on the pricing power the load-growth thesis depends on, just via a different mechanism than head-to-head rivalry. It answers the obvious question — if APS has a monopoly, why can it not simply charge more? — with: because the marginal large load is mobile before it sites, and a permanently cheaper alternative sits on the other side of a surveyed line. Management’s own risk factor concedes it: large loads “may locate their operations within service territories other than our own,” and the 10-K names “cooperatives, municipalities, electrical districts” as competitors.

Cutting the other way, and it matters: TSMC’s Phoenix complex is on APS, confirmed by the CEO on the Q1-2026 call.

Behind-the-meter self-supply. The 10-K names it directly as a competitive risk: “large customers developing large, utility scale generation to serve their energy needs.” This is the genuine long-term threat to the load-growth thesis — a hyperscaler that builds its own generation is not a ratepayer.

Distributed solar. Arizona’s solar resource is the best in the country and rooftop DG erodes the fixed-cost base; recovery via the LFCR is capped, lagged and subject to an earnings test. The Grid Access Charge was vacated by the Arizona Court of Appeals in mid-June 2026 on due-process grounds, with reconsideration motions pending and no mandate issued as at this report date.

Regulatory re-litigation. The most realistic threat to the franchise’s value is not entry but the periodic reopening of the compact itself — ROE, cost allocation, the fair-value increment, formula-rate legality. All four are live in the current docket.

Verdict (the competitive-position section). APS possesses the most absolute barrier to entry in the Greenwald framework — a legal monopoly — and, unusually, a genuinely advantaged asset within it in Palo Verde plus a genuinely advantaged return mechanism in the FERC transmission book. But on the test that matters, the franchise does not convert into an adequate return for the owner: APS has earned below its authorised ROE persistently, by 177bp in FY2025, and consolidated ROIC of 4.61% sits below a 7.63% requested WACC and below the 5.90% coupon on new 30-year paper. The moat protects the revenue base absolutely and the return on capital not at all, because the same authority that grants the barrier sets the return — and in Arizona that authority is elected, uses a historical test year, and applies a fair-value construct that cut the effective return to 4.39%. This is a durable franchise attached to an inadequate return, which is a materially different asset from a durable competitive advantage.

5. Growth History and Forward Opportunities

5.1 The demand record is genuinely excellent

There is no honest way to be dismissive about APS’s demand backdrop. It is among the best in the US regulated universe, and unlike most of the peer group it is delivered, not forecast:

*Q1-2025 excludes an $11m unbilled-revenue reduction; Q1-2026 is 9.4% before the prior-year adjustment.

That is nine consecutive quarters at or above the original 4–6% guidance range. Q1-2026 C&I growth was 14.6%. Residential customer growth ran 2.4% (2021), 2.2%, 2.1%, 2.3% and 2.5% (2025), guided 1.5–2.5% for 2026 — roughly double the national average per the Itron survey. The FY2025 system peak of 8,648 MW on 7 August 2025 was up 438 MW (+5.3%) on the prior year. Management guides 4–6% sales growth for 2026 and has raised the long-term range to 5–7% and extended it through 2030.

Phoenix-Mesa-Chandler added 59,065 residents in 2025 to reach 5,228,938, the fourth-largest absolute gain in the country. One caution: Maricopa County’s +35,411 represents roughly a 40% slowdown from the 2021–22 pace, on lower international migration.

5.2 The composition problem — this is the section that resolves the paradox

The obvious question is how a utility posts nine quarters of 4–7% sales growth and a negative five-year EPS CAGR. The answer is mix, and management has quantified it precisely. CFO Andrew Cooper, Q4-2025 call:

“1% of residential growth is somewhere north of $25 million, whereas 1% related to extra high load factor could be more in the $5 million to $10 million range.”

The 10-K’s own sensitivity is consistent: 1% of residential/small-C&I volume ≈ $25m of net income; 1% of large C&I ≈ $7m. A point of residential load is worth roughly 3–5x a point of data-centre load. This is the physics of load factor: an extra-high-load-factor customer draws a flat 24/7 profile that uses the fixed asset base efficiently and is therefore charged a much lower average price per kWh (business realises 13.19¢ against residential 17.03¢), while a residential customer’s peaky summer profile is what the whole system is sized for and is priced accordingly.

Management guides 5–7% long-term sales growth of which 4–6 points come from large C&I. Applying its own sensitivity, a 5–7% headline is worth roughly what a 1.5–2% residential-only growth rate would be worth. The market routinely conflates the two, and that conflation is a substantial part of the re-rating.

5.3 The large-load pipeline, stress-tested

APS discloses ~4.5 GW of committed load and just under 20 GW of uncommitted opportunity against an 8,648 MW system peak. Taken at face value the committed figure alone implies a >50% increase in peak demand; the uncommitted queue is 2.3x the entire system. Applying the CMS benchmark of $2–5bn of rate base per GW, the committed 4.5 GW alone could eventually imply $9–22bn of rate base against $12.5bn today — the committed queue could double the asset base.

That PNW nonetheless guides only 7–9% rate-base growth is itself the most revealing fact in the section: the bottleneck is not finding things to build. It is getting capital into rate base and earning on it.

Seven specific reasons to haircut the queue:

1. Management concedes double-counting. CEO Geisler, Q1-2026: “How much of that is duplicative projects or interest versus projects ready to execute is to be determined.”
2. Contract protections are undisclosed — the weakest disclosure in the peer set. APS has an extra-high-load-factor tariff it wants repriced ~45%, plus case-by-case “subscription” special contracts. No minimum-take percentage, no contract term, no collateral requirement and no exit fee has been disclosed on any of four earnings calls or in the 10-K. Compare: AEP discloses 63 GW with take-or-pay and minimum-demand charges, investment-grade/parent guarantees and ~$16bn of customer-funded interconnection; Southern discloses 11 GW fully contracted with minimum bills, collateral and cancellation fees; Evergy discloses 16–17 year terms with 5-year ramps and minimum monthly bills; Ameren discloses 12-year terms with 80%-of-capacity minimum demand. PNW discloses none of it.
3. Zero subscription contracts have been signed. The Tranche 1 subscription offering (1.0–1.2 GW) has been in “active negotiations” since roughly Q3-2025 and had produced no executed contracts as of the 4 May 2026 call. Geisler: “It is too early to tell how those may conclude.”
4. There is no ring-fencing — ratepayers hold the stranded-asset risk. Geisler, Q3-2025: “all of our investment goes in the rate base. The subscription model still goes into rate base.” This is the opposite of NiSource’s ring-fenced Genco structure. A large-load customer that walks away leaves a stranded asset for an elected commission to rule on.
5. Delivery is 2030 and beyond. Desert Sun at Gila Bend (up to 2,000 MW) has Phase 1 in late 2030 and Phase 2 in 2031–33, gated on the Transwestern Desert Southwest pipeline expansion (42″→48″, 1.5 Bcf/d) in 2029/30, to which APS has committed as anchor shipper ($7.3bn over 25 years of pipeline capacity). Red Hawk adds ~397 MW in 2028. A November 2025 all-source RFP seeks ≥1,000 MW for 2029–31.
6. Supply chain is a hard constraint. Turbine lead times have gone from 2–3 years in 2022 to ~5 years, with manufacturers now advising 7–8; prices are up roughly 300% in three years; GE Vernova is sold out through 2030, Siemens booked to FY2028, Mitsubishi to 2028.
7. The best local calibration is brutal. Tucson Electric Power disclosed roughly 8–10 GW of speculative interest against exactly one signed 286 MW contract (Project Blue, ACC-approved 3 December 2025) — a ~30:1 ratio between queue and contract in the same state, same year.

A defensible haircut: credit the 4.5 GW at face value, apply a 10–25% realisation rate to the ~20 GW, and the incremental figure is roughly 2–5 GW by the mid-2030s — about one-quarter of the narrative.

5.4 Which loads are actually in APS territory

This is the single most-confused fact in the bull case and it deserves stating plainly:

• TSMC Arizona (north Phoenix) — in APS territory. Phase 1 at ~200 MW, part of a programme the company has sized at ~$165bn. Geisler confirmed on the Q4-2025 call that “the 4.5 gigawatts does not include any potential expansion of TSMC” — genuine upside not in the number.
• Amkor’s $7bn packaging plant (raised from $5bn) — in APS territory.
• Intel’s Ocotillo/Chandler campus — in SRP territory, NOT APS. Salt River Project is building the 230kV campus substation under its High-Tech Interconnect Project. Intel’s ~$20bn expansion delivers nothing to PNW shareholders. SRP already serves 59 large-load customers totalling ~7,000 MW — a larger large-load book than APS’s entire system peak.
• Project Blue — in TEP territory, NOT APS.

Any analysis that credits APS with “Arizona’s semiconductor and data-centre boom” wholesale is materially overstating it.

5.5 Revenue growth decomposed

Headline revenue growth badly overstates economic growth because fuel is a pass-through that earns no return:

Revenue compounded at 8.3% over 2020–25 — but $940.0m of the $1,752.9m added (53.6%) was fuel and purchased-power pass-through. Margin revenue compounded at 5.6% over five years and 3.6% over ten. The headline is roughly half real.

5.6 Forward opportunities that are genuinely valuable

Three, and they are not the ones the narrative emphasises:

Transmission. The FERC-regulated book earns 10.75% on a formula rate with annual true-up, escaping ACC lag entirely. Transmission capex has gone from under $200m/yr five years ago to a $300–400m run-rate to ~$850m+ by 2028, with $2.6bn cumulative through 2028 and a $6bn+ FERC backlog through 2034. It contributed $0.16 of Q1-2026 EPS and the 2026 guide embeds ~$0.55. Roughly a third of incremental capital escapes Arizona lag entirely — the single highest-quality growth vector in the company.

Palo Verde licence renewal. Extending 1,146 MW of ~92%-capacity-factor carbon-free baseload from 2045–2047 to roughly 2065–2067 is a multi-decade rate-base annuity on an asset that cannot be replicated — nobody is building another 3.9 GW nuclear station in Arizona.

TSMC expansion, explicitly excluded from the 4.5 GW committed figure.

Verdict (the growth section). Low-quality growth — and the diagnosis is precise. The demand is genuinely best-in-class: nine consecutive quarters of 4–7% weather-normalised sales growth, 2.4% customer growth at twice the national rate, a record peak, and 4.5 GW of committed large load against an 8,648 MW system. The conversion into shareholder earnings has been approximately zero — diluted EPS fell from $5.47 (2021) to $5.05 (2025) and is guided to $4.65 in 2026. Three mechanisms explain it and all three are structural rather than cyclical: the growth arrives in the load class worth a quarter to a third as much per point; more than half of headline revenue growth is fuel pass-through earning nothing; and Arizona’s 30-month regulatory lag means the capital serving that growth earns nothing until a case is decided. Add the pipeline caveats — no disclosed take-or-pay terms, zero signed subscription contracts, no ring-fencing, 2030+ delivery, and a 30:1 queue-to-contract calibration next door at TEP — and the honest conclusion is that this is excellent demand attached to a poor earnings conversion mechanism. The growth is real; the claim on it is weak.

6. Financial Quality

6.1 The central fact: five years of capital, no earnings

Net income to common was $618.7m in 2021 and $616.5m in 2025 — down 0.4% — against $9.9bn of gross capex, a 29.4% increase in net plant and a 36.5% increase in total assets. Diluted EPS fell 7.7% and the entire decline is share issuance: absolute earnings were flat while the count rose 7.8%. Return on incremental invested capital over the period is approximately zero. FY2026 guidance of $4.55–$4.75 puts EPS 15.0% below the 2021 level.

6.2 The bridge — where the operating gain went

The operating business did grow. Every dollar of the gain, and more, was consumed below the line:

Two items did the damage. Interest expense rose 80.9% (+$188.7m), consuming 72% of the entire operating gain — the arithmetic consequence of funding a negative-FCF business with debt through a rate-rising cycle. And pension/OPEB non-service credits collapsed by $100.1m because prior-service credits became fully amortised on 31 January 2025. That second item is permanent and does not recur, with a further small drag in FY2026 (eleven months of credit in 2025 versus roughly none in 2026).

6.3 Earned versus authorised — the core diagnostic

APS earned a 7.78% GAAP ROE in FY2025 (APS net income to common $667.3m over average APS common equity of $8,577.4m) against a 9.55% authorisation — a 177bp shortfall worth ~$152m after tax, or ~$1.24 per diluted share.

Authorised ROE history: 8.70% (2019 case, decided November 2021, including a 20bp penalty later reversed to 8.90% via the Court Resolution Surcharge) → 9.55% (2022 case, rates effective March 2024) → 10.70% requested in the pending 2025 case, against ACC Staff at 9.55–9.80% and RUCO at 9.00–9.20%.

Five identifiable drivers, all structural:

1. Historical test year with 16–28 months of filing-to-rates lag. Roughly $2.0bn of rate base is outside rates as of the 2024 test year alone — ~$100m after tax, ~$0.82/share.
2. Capex outruns test-year recognition. FY2025 capex of $2,624.6m equals ~21% of the $12.5bn rate base added in one year.
3. The fair-value construct cut the effective rate of return to 4.39%.
4. The 2021 order was punitive — a sub-9% allowed return from December 2021 to March 2024 produced the trough years (2022 EPS $4.26, 2023 $4.41).
5. ~$90m/yr of unrecovered holdco interest.

Management concedes both gaps. CFO Cooper did not dispute an analyst’s “roughly 200 basis point” characterisation of the EPS-versus-rate-base delta, answering that “we will have to revisit all of this at the conclusion of the rate case.” CEO Geisler on the earned gap:

“Management’s goal is to be able to consistently earn within that 50 basis points, given there is some element of structural lag that will continue to exist… would allow us to do so by 2029 and going forward.”

Management’s own best case is to still be under-earning by 50bp, and not before 2029. And the historical delta has been far worse than 200bp: net plant compounded 6.66%/yr over 2021–25 against diluted EPS at −1.98%/yr — an 864bp gap. Over 2015–25 it was 5.6% versus 2.6%, a 300bp gap. Treat “200bp” as an aspiration, not a track record.

The double-leverage paradox. Consolidated ROE (8.93%) exceeds APS’s (7.78%) because $1,915.4m of holdco debt is pushed down into APS as equity: APS’s GAAP equity is $8,881.6m while PNW consolidated equity is $7,046.5m, and the ~$1.84bn difference is the holdco debt. Parent net interest of ~$90m/yr (~$68m after tax) equals ~99bp of consolidated ROE that is structurally unrecoverable in rates, and it is why PNW’s consolidated debt/capitalisation (~60%) far exceeds APS’s (~50%).

6.4 Margins — half optical, half real

Reported operating margin fell from 25.5% (2017) to 20.0% (2025). Normalising for the fuel pass-through that inflates the denominator:

The 5.5-point reported compression decomposes as roughly 3.0 points of fuel-mix optics (55%) and 2.5 points of real compression (45%). The real half broke in 2018–19 — dating to the 2017 rate order and the punitive 2019 decision, not to operating deterioration — and has since recovered from a 27.2% trough (2022) to 31.3% as the 2022-case rates phased in, though it remains ~4 points below the 2015–17 level. Note that the aggregator’s “gross margin 41.6%” is not a utility-meaningful figure; revenue less fuel ($3,406.5m, 63.8% of revenue) is the right base.

6.5 Quality of earnings

The tax rate is low but the usual “cliff” narrative is wrong. The consolidated effective rate was 14.45% (2025), 15.01% (2024) and 12.91% (2023), against 33.7–34.3% pre-TCJA. FY2025 reconciling items totalled −$73.7m (−9.98% of pretax, ~$0.60/diluted share): excess deferred income taxes −$36.6m, solar/wind PTC −$14.7m, ITC amortisation −$12.6m, AFUDC-equity −$7.0m, other regulatory amortisation −$2.8m. There is no near-term cliff — EDIT amortisation has been a flat $36.6m in each of 2023, 2024 and 2025 against regulatory liabilities of $847.6m (ACC, through 2046) and $200.2m (FERC, through 2058), roughly twenty more years of runway. More importantly, EDIT is a ratemaking flow-back: it lowers tax expense and the revenue requirement roughly in step, so it is closer to EPS-neutral than the headline suggests. Any model that grosses FY2025 EPS up at a “normal” utility tax rate is wrong in both directions.

A hidden two-sided item: the nuclear PTC. APS claimed a $33.4m nuclear production tax credit on its 2024 return and intends to claim $39.6m for 2025, using a revenue-requirement methodology for “gross receipts from nuclear sales.” Absent IRS guidance the entire benefit is reserved as an uncertain tax position and no income tax benefit has been recognised — unrecognised tax benefits jumped from $44.2m to $105.5m in 2025. $73.0m of cumulative 2024–25 cash-tax benefit (~$0.60/share) sits off the P&L: it releases to earnings if the IRS accepts the methodology and is repayable if not.

AFUDC is a large and rising share of “earnings.” Allowance for funds used during construction — a non-cash capitalised return that becomes cash only when plant enters service and the ACC puts it in rates:

Roughly one dollar in six of FY2025 earnings is AFUDC, and the $22m year-on-year increase in AFUDC-equity more than accounts for the entire $8m earnings improvement. With CWIP at $1,649.5m and capex stepping to $2.7bn, this share rises rather than falls.

One-time items. FY2024 contained a $22,988k gain on the Bright Canyon sale (plus $6,423k in 2023) — FY2024 EPS of $5.24 is flattered by roughly $0.15. FY2025 contains the $198.7m Palo Verde sale-leaseback buy-in and the Cholla retirement. On pensions the cash position is benign: zero required or expected contributions for 2026, 2027 and 2028, none made in 2024 or 2025, and the $723.0m pension regulatory asset earns a full return in rate base.

The accumulated-depreciation puzzle, resolved. Accumulated depreciation fell from $9,412.2m to $9,234.0m in 2025 despite $969.6m of D&A. This is the Cholla retirement — roughly $750m of gross plant charged to accumulated depreciation under the standard convention, with no P&L effect, no impairment and no securitisation. Only $81m of net book value remained and it continues to earn a return in rates. Rate base and ROE are essentially undistorted. The open item is that APS withdrew its standalone deferral application on 8 July 2025 and rolled Cholla recovery entirely into the 2025 Rate Case.

6.6 Balance sheet, leases and the maturity wall

Debt, correctly decomposed. The frequently-quoted “$14.3bn of debt, up 29%” is misleading. Reconciling to the balance sheet: long-term debt less current maturities $9,205.7m + current maturities $600.0m + short-term borrowings $757.0m = true borrowings of $10,562.7m, plus operating lease liabilities of $3,737.0m = $14,299.6m. These are operating leases, not finance leases. So the one-year move decomposes as true borrowings +$1,135.6m (+12.0%) and operating leases +$2,115.7m (+130.5%) — roughly 65% of the apparent spike is lease capitalisation.

What the leases are, and why they matter anyway. Nine energy-storage PPA operating leases commenced in 2025 (+$2,195.7m of right-of-use assets), on 20-year terms with exclusive use and fixed capacity payments, and the 10-K states they are recoverable through the PSA. The number that matters is forward: $11.4bn of executed-but-not-yet-commenced leases, primarily storage, commencing between April 2026 and June 2028 — already landing at +$410.6m in Q1-2026 alone. That is a figure approaching the entire current rate base. Economically these are PSA-recoverable resource commitments rather than financing of an equity shortfall, but they will balloon the balance sheet. A tell: on 18 February 2026 both credit agreements were amended to exclude PSA-recoverable PPAs and storage leases from the covenant definition of indebtedness. Management negotiated a carve-out for exactly the obligation that is exploding; banks granted it, and rating agencies generally do not grant the equivalent. In context, PNW’s leases equal 35.4% of borrowings against a peer median of 0.4% — an order-of-magnitude outlier, though part of that is GAAP classification since peers’ PPAs often fail the control test.

Leverage in peer context — mid-pack, not an outlier. PNW’s consolidated debt/capitalisation of 60.0% compares to a peer median of 59.4%, ranking 7th of 13. Including capitalised leases it moves to 3rd at 67.0%. Both PNW and APS are capped at 65% consolidated debt/total capitalisation under their credit agreements; at 31 December 2025 PNW was ~60% and APS ~50%, leaving roughly five points of covenant headroom. The trend is nonetheless adverse: net debt/equity rose from 111.7% (2020) to 148.9% (2025), and the tangible common equity ratio fell every single year, from 27.71% to 22.10%.

The maturity wall is concentrated at the holdco.

The 2026–29 wall is $2,394.0m, of which 76% sits at the holding company — where there is no rate base and no recovery mechanism. 2027 is heaviest and includes the $525m convertible. Note that Fitch rates PNW holdco commercial paper F3, the lowest investment-grade short-term tier, materially narrowing the parent’s CP investor base precisely where those maturities sit.

Refinancing cost is already visible. Long-term debt carries at $9,805.7m against a fair value of $9,164.5m — the book trades 6.5% below par (APS’s 8.7% below) because existing coupons are far below market. New money is being raised at 4.90–5.90%: PNW issued $400m of 4.90% 2028s and $400m of 5.15% 2030s in May 2025 to retire $500m of 1.30% notes; APS issued $700m of 5.90% 2055s and reopened $250m of 5.70% 2034s in August 2025. Every refinancing is a 250–450bp coupon step-up. Interest charges rose from $254.3m (2021) to $469.7m (2025), +84.7%, and Q1-2026 charges of $125.8m versus $104.9m (+19.9%) imply an annualised run-rate above $500m.

The ratings arc is itself evidence. Between mid-2021 and the November 2021 order, all three agencies acted: Fitch downgraded PNW and APS on 12 October 2021, Moody’s followed on 17 November and S&P on 9 November. Pinnacle West went from Moody’s A3 / S&P A− / Fitch A− before the decision to Baa2 / BBB+ / BBB today — two notches lower at Moody’s and Fitch, one at S&P — and it has recovered none of it, despite the constructive February 2024 order. The agencies’ outlooks normalised to Stable within five weeks of that order; the ratings did not. Note also that Fitch’s March 2024 downgrade of the parent (while affirming APS) was attributed to holdco leverage and structural subordination rather than Arizona regulation — independent corroboration of the double-leverage problem set out in the competitive-position section. The cost was quantified in APS’s own rate-case testimony: it was unable to place the full $500m of debt it sought at a competitive rate and had to settle for $400m at what was then the highest coupon in its portfolio.

Ratings and cushion. As of 20 February 2026: APS at Moody’s Baa1 / S&P BBB+ / Fitch BBB+ and PNW at Moody’s Baa2 / S&P BBB+ / Fitch BBB, all Stable and all reaffirmed in April 2026. Management targets PNW FFO/debt of 14–16%, stating the midpoint represents “>100bps cushion above Moody’s threshold.” On the Q4-2025 call the CFO was more specific: “Moody’s is really the limiting constraint given their downward threshold is 14%… we’re high 14s… our aspirational goal is 100 basis points of cushion” — i.e. the actual cushion is roughly 50–80bp, below management’s own stated aspiration. Holdco debt is ~18% of total against a “mid-teens” target. An independent FFO/debt proxy computes to ~14.7% on true borrowings and ~10.8% including capitalised leases, both before the agencies’ own adjustments, which would push the figure lower rather than higher. This is why the pending rate case is a ratings event, not merely an earnings event.

6.7 Cash flow: free cash flow has been negative every year

Operating cash flow covered 68.8% of FY2025 gross capex (67.9% over five years). Every dollar of the $1,952.4m of dividends paid over five years was funded with external capital. The $4.09bn gap was closed with net new long-term debt of $3,510.1m, short-term borrowings, and roughly $0.5bn of equity — i.e. ~92% of the external capital that funded both the business and the dividend came from debt.

One genuine tailwind that is now spent: the deferred fuel regulatory asset unwound from $463.2m (2023) to $287.6m (2024) to $149.1m (2025), collecting $463.0m from customers in 2025. The PSA under-recovery has largely reversed, and that support to 2023–25 operating cash flow does not repeat.

Verdict (the financial-quality section). No — the economics do not improve with scale, and on the two tests that actually matter for a regulated utility the company fails both. Does it earn its authorised return? No: APS earned 7.78% against 9.55% in FY2025, a 177bp / ~$152m / ~$1.24-per-share shortfall, with management’s own target being to close that only to within 50bp and only by 2029. Does it convert capex into shareholder value? No: $9.9bn of gross capex over five years produced net income to common that was lower in 2025 than in 2021, with the entire operating gain consumed by an 80.9% rise in interest expense and the permanent loss of $100m of pension credits. Free cash flow was negative in all five years (−$2.1bn after net capex, −$4.1bn after dividends); 100% of the dividend is externally funded; ROIC of 4.61% sits against a 7.63% requested WACC and a 5.90% coupon on new 30-year paper. Roughly one dollar in six of reported earnings is non-cash AFUDC. What genuinely does improve with scale is narrow but real: the FERC formula-rate transmission book, which has contemporaneous recovery, an annual true-up and is the fastest-growing capex bucket. Everything else is a capital-intensive business compounding rate base at 6–7% and shareholders’ earnings at approximately zero.

7. Capital Allocation

7.1 The headline test

Over FY2021–FY2025, capital employed (total debt excluding leases plus common equity) rose from $13,262m to $17,609m — +$4,347m, or +32.8% — and total assets rose 36.5%. Net income to common went from $618.7m to $616.5m. The incremental return on $4.35bn of incremental capital is approximately zero. ROIC fell from 5.05% to 4.61%; ROE from 10.72% to 8.93%. This is the fact against which every capital-allocation judgment in this section must be measured.

7.2 The crux: is issuing equity at 1.8x book into sub-authorised rate base value-creative?

This is the central capital-allocation question for PNW and it deserves both sides.

Book value per share at 31 December 2025 was $58.28 ($7,046.5m over 120.9m shares); at $106.06 that is 1.82x book.

The case for. A dollar of equity raised at 1.82x book is infused into APS at 1:1 and becomes a dollar of rate-base equity earning ~8–9%. Because the buyer paid $1.82 for $1.00 of book, the new share carries roughly 1.82x the earnings contribution of the average incumbent share — mechanically EPS- and BVPS-accretive, and this is the entire basis of the utility growth machine. The market currently capitalises those rate-base earnings at ~22.8x (a 4.39% earnings yield) against an ~8.9% book return, so each $1.00 of equity raised is instantly marked at roughly $2.03 of market value.

The case against, which is the one that matters. Value is created only where the earned return exceeds the cost of capital. APS earns a 7.78% GAAP ROE. A regulated electric utility’s cost of equity at a 0.4–0.6 beta with a ~4–4.5% long bond is roughly 7.5–9%. The spread is zero to negative, inside the error bars at best. On invested capital it is worse: 4.61% book ROIC against an ~6% WACC and a 7.63% requested WACC. PNW is deploying $2.6bn a year at a return that does not demonstrably clear its cost of capital.

The empirical adjudication settles it. If the capex were genuinely value-creating, EPS would have grown. It did not. Rate base, assets, debt and share count all grew; earnings per share fell. The economic surplus from the buildout accrued to customers (via lagged and disallowed recovery) and to bondholders (via a bigger, better-covered asset base), not to the residual claimant.

The honest verdict is neither of the crude versions: issuing above book into rate base is arithmetically better than issuing below it, so this is not value destruction in the simple sense. But it is not value creation either — it is a multiple-preservation trade that works only so long as the market keeps paying ~22.8x for an 8.9%-ROE, sub-authorised, regulatory-lagged earnings stream. The moment the multiple normalises toward the utility mean, the mechanism reverses and the dilution becomes real. That is the single most important capital-allocation fact about this company, and it is precisely why the current 98.5th-percentile valuation is a risk to the business plan and not merely to the share price.

7.3 The equity ledger — including one large, badly-timed raise

2021–2023: essentially no equity issued. Net common stock issuance was negative in each year (−$2.4m, −$2.7m, −$4.1m) and weighted-average diluted shares were flat at 113.2m → 113.8m.

February 2024 — the block forward, struck at the worst price of the cycle. On 28–29 February 2024 PNW entered forward sale agreements on 11,240,601 shares — roughly 9.9% of the then-outstanding count — at an initial forward price of $64.505, an aggregate $725.1m. That was ~1.18x the then-book value of ~$54.7 and near the multi-year price low. Settlements to date: 5,377,115 shares at $64.17 (December 2024, $345.0m); 243,186 at $63.12 (September 2025, $15.4m); 1,193,950 at $62.82 (December 2025, $75.0m). 4,426,350 shares remain undelivered, with maturity extended — twice — most recently to 31 December 2026.

Measured against today’s $106.06, the ex-post value foregone across the full 11.24m shares is roughly $41.55 × 11.24m ≈ $467m. That is a hindsight calculation and should be labelled as such, but the size and the strike are facts, and the decision to twice extend settlement rather than deliver is an implicit acknowledgement that the price was poor.

November 2024 onward — the ATM, accelerating. A $900m at-the-market programme with forward-sale capability opened 8 November 2024. At 31 December 2025 there were four outstanding forwards on 2,199,415 shares at a weighted-average $90.10. Q1-2026 alone added roughly $264m of new forwards at $98–101, taking the total to nine forwards on 4,861,922 shares at a weighted-average $95.08, with a tenth added on 1 April 2026. By 5 June 2026, ~$630m of the $900m programme had been used, leaving only ~$270m against a $7.95bn three-year capex plan. A June 2026 amendment extended the forward outside-maturity from 18 to 24 months.

To management’s credit, the 2026 forwards are struck at $98–101 versus the February 2024 block at $64.51: the blended cost of equity capital has improved materially as the stock re-rated. But a new or upsized authorisation is a near-certainty, and on 21 May 2025 shareholders approved doubling authorised common stock from 150,000,000 to 300,000,000 shares — not a housekeeping item, but capacity for a materially larger programme.

The convertible. $525m of 4.75% Convertible Senior Notes due 15 June 2027, priced June 2024 at a conversion price of ~$92.30 (a ~20% premium to the then-$77 stock). Principal settles in cash, with only the excess in cash or shares at PNW’s option. At $106.06 the notes are ~15% in the money, implying roughly $78m of embedded dilution or cash cost maturing inside the heavy 2027 wall. There are no hybrids, no junior subordinated notes and no preferred stock.

A real governor worth noting: the ACC’s financing order of 17 December 2024 caps annual equity infusions into APS at 2.5% of APS’s total assets (~$725m/yr) on a three-year rolling average, subject to APS’s equity ratio staying below the last approved capital structure plus 50bp. Pinnacle West contributed $300m into APS on 15 May 2025.

7.4 The dividend — externally funded, and growth ratcheted to a token

Dividends per share paid: $3.34 (2021), $3.42, $3.48, $3.54, $3.60 (2025). The indicated rate rose by exactly $0.06 every year for five years, which produces a mechanically decelerating growth rate: +2.41%, +1.76%, +1.73%, +1.70%, +1.68%. The Q2-2026 declared rate is $0.91/share. Payout ratios computed from the filings: 59.7% (2021), 78.3%, 77.1%, 64.8%, 68.6% (2025) — a five-year cumulative 69.0%. On 2026 guidance (~$3.70 DPS against ~$4.65 EPS) the payout rises to roughly 80%.

Two things must be said plainly. First, no numeric payout-ratio target appears anywhere in the 10-K. The only stated policy is that the dividend depends on board declaration and “a number of factors, including our financial condition, payout ratio, free cash flow and other factors.” The “sustained 65–75%” figure is an investor-deck construct, not a filed commitment — and 2026 guidance already breaches it.

Second, PNW has not earned a dollar of its dividend in cash in five years. Cumulative post-capex FCF was −$2,135m against $1,952m of dividends paid, with the $199m Palo Verde buy-in and ~$103m of NCI distributions on top: a total external funding requirement of roughly $4.4bn, of which approximately 92% came from debt. This is not unusual for a growth-capex utility and it is not a solvency observation. But it should be stated without euphemism: the yield is a return of capital raised from the capital markets, not a distribution of cash the business generated.

Buybacks: none. There is no repurchase programme, no issuer-purchases table in Item 5, and no authorisation anywhere in the five-year 8-K corpus. PNW is a structural net issuer.

7.5 M&A and non-utility ventures — a poor record, now largely closed

The record is short and unimpressive. Bright Canyon Energy was sold in 2024 (~$84.3m of proceeds; a $22,988k gain in FY2024 and $6,423k in FY2023). The 9.9% Clear Creek wind investment was fully impaired in Q4 2022. What remains in PNW Power is a 50/50 TransCanyon joint venture with Berkshire Hathaway Energy pursuing the 214-mile Cross-Tie transmission line (BLM Record of Decision December 2025) and a 5.1% stake in Nobles 2. El Dorado holds a ~$21m equity-method position in SAI Advanced Power Solutions, a switchgear manufacturer serving data centres — a position management describes as “not something that we plan for when we budget.”

The most significant recent transaction is not M&A but a buy-in: the ~$198.7m acquisition in September 2025 of two of three Palo Verde Unit 2 sale-leaseback interests (~94 MW), which reduced non-controlling interests from $103.2m to $40.6m and created a $151.5m regulatory asset with no stated amortisation period — recovery of which is not yet approved. On the merits this is a sensible transaction: buying back a slice of the single best asset in the fleet, at a moment when firm carbon-free capacity is scarce, and putting it into rate base. It is the clearest example of good capital allocation in the file.

7.6 Compensation — the mechanical explanation for the behaviour

This is where the section’s verdict is actually decided, because the incentive design explains the pattern of results precisely.

Annual cash incentive (2026 proxy): 50% PNW/APS Earnings Performance + 50% Business Unit Performance, with an earnings threshold gating any payout. The business-unit half is 70% shared enterprise metrics, comprising Employee Safety 15%, Customer Experience 20%, Customer Affordability 35% — split into “PNW O&M” 20% (result 0) and “PNW CAPITAL” 15%, measured as capital spend against “External Target ±1.00%,” which paid 200% — and Reliability 30% (result 173%).

Long-term incentive: 70% performance shares / 30% RSUs over three years, weighted 40% relative TSR versus the EEI Index, 40% “EPS Performance… actual EPS to our EPS target” (an internally set target), and 20% “MW Installed.” The 2026 grants shift to Relative TSR 34% / EPS 33% / “Reliable and Affordable” 33%.

There is no earned-ROE metric, no authorised-return-achievement metric, no ROIC metric and no regulatory-lag metric anywhere in either plan. Management is paid on absolute earnings dollars against a committee-set target, EPS against an internally set target, megawatts installed, and capital spend versus budget. Every financial metric is satisfied by growing the asset base and spending the plan; none requires actually earning the authorised return on that base. That is a direct, mechanical explanation of the capital-allocation section result: assets +36.5%, earnings flat.

The most damning specific, verbatim from the 2026 proxy: “For 2025, we set the APS earnings target at $622 million, a decrease to 2024 actual incentive earnings of $641.9 million. Likewise, we set Pinnacle West’s 2025 earnings target at $544 million, a decrease to Pinnacle West’s 2024 actual incentive earnings of $608.8 million.” Actual 2025 PNW incentive earnings were $594.7m — still below 2024’s $608.8m — and paid at 182% of target; APS paid 173%. Executives earned near-maximum on the earnings half of the bonus for delivering lower earnings than the prior year, because the committee set the bar below the prior year’s actual. The committee’s stated rationale — that 2024 was weather-flattered — is defensible on its own terms, but the outcome is a plan calibrated so that a down year pays 1.8x target.

A TSR reality check. On the 10-K’s own performance graph, $100 invested at 31 December 2020 was worth $138 (PNW) against $144 (EEI Index) and $195 (S&P 500) at 31 December 2025 — PNW underperformed its own sector index over five years. Yet the 2022 performance-share grant paid 186.5% of target on an “actual TSR percentile performance of 83.7%,” and the 2023 grant paid 159%. The metric is the right one; rolling three-year windows measured off the depressed 2021–22 base flattered it.

Pay levels and governance. CEO Guldner: $9.322m (2023), $10.974m (2024), $2.828m (2025 partial). CEO Geisler: $3.255m (2023), $3.865m (2024), $8.415m (2025). CFO Cooper: $2.654m / $3.126m / $3.393m. Pay ratio 61:1 / 65:1 / 42:1. Say-on-pay support has been 94.8%, 94.8%, 96.6% and 97.3% at the May 2026 meeting — no shareholder pressure whatsoever on plan design. Governance is otherwise clean: single share class, annual terms, independent lead director, double-trigger change-of-control, no excise gross-ups, clawback in force. One irritant: director Glynis Bryan has drawn persistent ~5.7% withhold votes following a Glass Lewis “overboarded” recommendation. One recent item worth flagging: on 23 June 2026 the HR Committee granted CNO Adam Heflin a $1.5m off-plan “Discretionary Credit Award” into his deferred compensation account, cliff-vesting 1 May 2030, with no performance condition — a pure retention payment.

7.7 Insider behaviour — near-total absence of conviction

Parsing all 202 Form 4s and 10 Form 144s in the five-year corpus (1,141 individual transaction lines):

• Open-market purchases (code P) in five years: exactly one. Director Susan T. Flanagan, 500 shares at $86.54 on 5 August 2024 — $43,270. That is the entire universe of insider buying. No officer — not the CEO, CFO or COO — has bought a single share on the open market in five years.
• Sales (code S): 19 transactions, 47,283 shares, $4.09m gross. Dollars sold to dollars bought is 72:1.
• Not one Form 4 references a 10b5-1 plan. Of the 129 filings carrying the post-2023 checkbox, all 129 report no plan. Every sale is discretionary.
• By person: COO Tetlow sold 24,809 shares — by far the largest seller — and holds just 5,160 today, including 6,567 shares at $99 on 13 May 2026, in the middle of the rate-case hearing. Guldner took 166,500 shares of grants with zero sales and zero purchases; Geisler 62,366 grants, no purchases.
• Routine flow was 292 code-F tax-withholding events (234,209 shares, $20.3m) and 162 code-D dispositions to the issuer. Code-G “gifts” are transfers into family trusts and net to zero.
• Directors and executive officers as a group (20 people) own 229,489 shares — 0.19% of the company, ~$24.3m. CEO Geisler holds 37,567 shares (~$4.0m) against a 5x-salary guideline on a $1.0m salary. The register is dominated by index funds: Vanguard 12.60%, Capital Research 11.30%, BlackRock 8.20%, State Street 6.02%, and Barrow Hanley 5.58% — the only active value holder of size.

With the stock at a 98.5th-percentile own-history valuation, insiders bought $43,270 and sold $4.09m. There is no conviction signal in either direction from officers, and the single director purchase is too small to carry meaning. The honest read is that management has essentially no personal economic exposure to the outcome of the capital plan it is executing.

Verdict (the capital-allocation section). No — this is competent operational execution of a plan that has not paid shareholders. The team has run the plant well (a 92% Palo Verde capacity factor and a 100% summer factor), held core O&M flat against a rapidly growing customer base, bought back a slice of its best asset in the Palo Verde lease buy-in, exited a failed non-utility diversification, and materially improved the cost of its equity issuance between 2024 and 2026. But the capital-allocation record is poor on the measure that counts: $4.35bn of incremental capital employed produced zero incremental earnings; equity was sold forward at $64.51 in a raise representing ~9.9% of the company near the cycle low; 100% of a five-year dividend was funded externally, ~92% of it with debt; there are no buybacks and the company is a structural net issuer; and the payout is heading to ~80% on 2026 guidance against a target that exists only in an investor deck. Most tellingly, the incentive plan contains no earned-return, ROE or ROIC metric of any kind — it pays for megawatts installed, capital spent against budget, and earnings against a target the committee reset downward, which paid 182% for a year in which earnings fell. Management is doing exactly what it is paid to do. That is the problem, and it will not change until the metrics do.

8. Changes and Headwinds — Last Two Years

The two-year record is a study in divergence: an exceptional demand outcome and a deteriorating earnings outcome, simultaneously.

Verdict (the recent-changes section). Neutral-to-negative, and the balance has deteriorated over the last eight months. The genuine positives are concentrated in legislation rather than regulation — H.B. 2201’s wildfire safe harbour is a material structural improvement and H.B. 2679’s securitisation authority is useful — and in the operating record, where demand outcomes have been outstanding and the Palo Verde buy-in was a good trade. Against that: the ACC denied the wildfire deferral, capped equity infusions, cut the DSM budget, and has not yet delivered the formula rate on which the entire “5–7% becomes evergreen” thesis rests — a mechanism that became sub judice in November 2025. Earnings have gone $5.24 → $5.05 → a guided $4.65 through the best demand environment in the company’s history, the authorised-share count has been doubled, the ATM is ~70% consumed, and the payout is heading toward 80%. Everything now rests on one binary decision, before a commission that has ruled against APS before and whose chairman was repudiated by his own party’s voters four days ago.

9. Risk Analysis

The three that matter. Risks 1, 2 and 3 are not independent — they are the same risk viewed from three angles, and they resolve on a single date. An elected commission facing a general election votes, in December 2026, on a 14.75% rate increase, a 115bp ROE increase, and a formula-rate mechanism whose legal basis is simultaneously before the state Supreme Court. This is a genuinely binary event and it is four to five months away. The most important asymmetry, established in the price-action section, is that the market has historically priced Arizona regulatory outcomes as one-directional: the 2021 adverse order cost 8% in a day; the 2024 favourable order was worth nothing.

10. Valuation Discussion — Embedded Expectations

This section discusses valuation only as embedded expectations and scenario analysis. It contains no price target and no recommendation.

10.1 Where the stock trades

10.2 The own-history read — and the sector context that qualifies it

On the AZI valuation index, PNW’s composite own-history percentile is 98.5 against roughly a decade of its own range: P/E at the 97.6th percentile, P/B at the 99.3rd, P/S at the 98.6th. All three are at or near record simultaneously.

Its own decade of year-end prints puts today in context:

Today’s 1.82x book is the top of the decade range and ~21% above the 2025 year-end mark. The entire move is multiple expansion, not book growth. Enterprise value grew from $16.26bn (2021) to $24.95bn (2025), +53%, while diluted EPS fell from $5.47 to $5.05 — EV growth has come from debt, not earnings.

But the sector context matters and must be given its weight. The entire regulated-utility group sits at extreme own-history percentiles: OGE 99.98, EVRG 99.90, LNT ~99.8, NI ~99, PNW 98.5, IDA 99.2, WEC 95.7, DTE ~95.5, CNP ~95, ETR 94.7, ATO 94.0, XEL 93.0, EXC 93, AEE 92.4, ED ~92, DUK ~80, FE ~80, CMS ~78, D ~72, EIX 64.4, NEE 59. PNW ranks roughly third-richest of sixteen against its own history — but the re-rating is a sector event driven by the AI-power narrative and the rate cycle, not a PNW-specific mania. Anyone treating the 98.5th percentile as evidence of a company-specific bubble has mistaken a sector fact for a stock fact.

10.3 Cross-sectional: the discount that isn’t quite what it looks

Against the regulated peer set on the same date and source:

On trailing figures PNW looks cheap: a 14.3% P/E discount and a 19.1% P/B discount to the group, while earning a median ROE (9.39% versus 9.46%). That framing invites the conclusion that the discount is unearned.

It does not survive the forward test. Every peer in the set guides to positive mid-to-high single-digit EPS growth (NI 9–10%, AEP >9%, NEE 8%+, ETR 8–9%, CNP 7–9%, LNT ~7%+, ATO/CMS/DTE/FE 6–8%, EXC/DUK/D/EIX 5–7%, ED 5–6%). PNW’s FY2026 EPS is guided down ~8%. On the year the company is actually guiding to, PNW’s 22.8x is roughly in line with the peer median of 23.08x on rising earnings. The trailing discount is an artefact of the denominator, not a valuation gap.

The correct conclusion is therefore neither “cheap versus peers” nor “expensive versus history” — both are true and neither settles it. PNW is at parity with the group on forward earnings while carrying single-state concentration, an elected regulator, an RRA Below Average jurisdiction, a pending 14.75% rate request, a five-year record of underperforming its own sector index, and the thinnest rating-agency cushion in the set. That is not a discount for risk; it is at best neutral pricing for demonstrably higher risk.

10.4 What the current price is underwriting

At $106.06 on FY2026 guidance of $4.65, the market pays 22.8x for a company whose EPS has compounded at −2.0% over five years. For that multiple to be defensible, several things must be true simultaneously:

1. FY2026 is a trough, not a trend. The market is looking through a guided-down year to a post-rate-case recovery.
2. The ACC materially closes the earned-versus-authorised gap. Theoretical maximum uplift: closing APS’s 7.78% earned ROE to the 9.55% authorisation is worth +$157m, or +$1.29/share; the requested step from 9.55% to 10.70% on ACC equity adds +$75m, or +$0.62/share. Full realisation implies a run-rate near $6.56 versus the $4.65 guided — a 41% uplift. The market is plainly capitalising a substantial fraction of that.
3. The FRAM is approved and survives judicial review, converting a 30-month lag into an annual true-up.
4. The large-load pipeline converts on something like the disclosed terms.
5. The multiple holds, because the funding model itself depends on it (the capital-allocation section).

Points 2 and 3 must clear an elected commission in December 2026. Point 3 must also clear the Arizona Supreme Court. These are not one bet; they are four, and they are not independent.

10.5 Scenario analysis

Assumptions stated explicitly. EPS figures are FY2028 run-rate, since rates from the pending case take effect in early 2027 and a formula-rate first adjustment would not produce a full year until 2028. Values are discounted two years at 8% and two years of dividends (~$3.70/yr) added.

Probability-weighted: ~$91, against a spot price of $106.06.

Multiple assumptions are anchored on PNW’s own decade (year-end P/E range 12.9x–19.7x, median ~17.2x), widened at the top end to reflect that a genuine formula rate would justify a structurally higher multiple than the company has historically earned.

Cross-checks, independently derived:

• Price-to-book. BVPS was $58.28 at 31 December 2025, compounding at ~4.5%: ~$60.90 (2026E), ~$63.64 (2027E). At the decade-median 1.50x, 2027 book implies ~$95; a 1.45x–1.65x band implies $92–105. Today’s 1.82x is the decade high.
• Dividend yield. An indicated $3.70 at PNW’s own historical 3.8–4.2% band implies $88–97. Today’s 3.5% is at the rich end of the decade.

All three approaches — scenario-weighted (~$91), price-to-book ($92–105 on 2027 book, ~$95 at the median multiple) and dividend yield ($88–97) — cluster in the high-$80s to low-$100s, below the current price. The convergence is worth noting precisely because the methods are independent.

10.6 What the market is pricing correctly, and what it may not be

Correctly priced: the demand backdrop (it is real, delivered and best-in-class); the value of Palo Verde and of a licence extension into the 2060s; the FERC transmission book’s contemporaneous recovery; the sector-wide rate-cut and AI-power bid; and the fact that a formula rate would be genuinely transformational.

Possibly mispriced, in the bulls’ favour: the H.B. 2201 wildfire safe harbour is a material reduction in tail risk versus the western peer group and receives almost no discussion; the FERC transmission mix shift is under-modelled; and the $73.0m of reserved nuclear PTC is a real off-P&L asset.

Possibly mispriced, against the bulls: the load-quality discount (a point of data-centre load is worth a quarter to a third of a point of residential load — this is quantified by management and largely absent from the narrative); the fact that the FRAM is sub judice; the elected commission and the 21 July 2026 primary result; the absence of any disclosed take-or-pay protections on 4.5 GW of “committed” load; that all of it enters rate base with no ring-fencing; and the historical asymmetry in how this market prices Arizona orders.

11. Variant Perception

11.1 Consensus

The consensus view is straightforward and not unreasonable: Pinnacle West owns the best-growing service territory in the American utility sector; the 2025 Rate Case plus a formula rate will finally fix the regulatory lag that has held earnings back; the AI/data-centre buildout in Phoenix gives it a multi-decade rate-base runway; and the stock still trades at a discount to peers. Comparable analyses of fifteen other regulated utilities landed on a hold-type conclusion for every one of them, with the sector-wide framing being “own the algorithm, not the multiple.” PNW is being bought as a late convert to that algorithm.

11.2 The strongest bull case

Stated at its strongest, because it deserves to be:

APS has 4.5 GW of committed load against an 8,648 MW peak, and applying the industry benchmark of $2–5bn of rate base per GW, the committed queue alone could double a $12.5bn rate base. Nine consecutive quarters of 4–7% weather-normalised sales growth is delivered performance, not a forecast, and TSMC expansion is explicitly excluded from the 4.5 GW. The company is chronically under-earning by 177bp — which means there is $1.29 per share of earnings sitting on the table that requires no new capital, only a regulatory decision, plus another $0.62 if the ROE step is granted. Both APS and ACC Staff have now put formula-rate parameters on the record, and UNS Gas received one in February 2026 at ~86% of its ask with a formula “for perpetuity.” Arizona has legislated a wildfire safe harbour that abolishes inverse-condemnation claims — structurally better than California — and securitisation authority. The FERC transmission book earns 10.75% on a formula rate free of state lag and is growing fastest. APS rates have risen 2.67%/yr against 3.62% CPI, giving real headroom for an increase. And the stock trades at a 14% P/E and 19% P/B discount to peers while earning a median ROE. If the December order delivers, this is a company whose earnings step up 20–40% and whose growth algorithm becomes credible for the first time in a decade.

11.3 The strongest bear case

Also at its strongest:

PNW has had the best demand backdrop in the sector for five years and delivered a −2.0% EPS CAGR. That is not bad luck; it is structure. Arizona’s historical test year imposes a 30-month lag on a $2.6bn/yr programme; the fair-value construct cut the effective return to 4.39%; ~$1.9bn of holdco debt earns nothing and costs ~99bp of consolidated ROE; and the growth arrives in the load class management says is worth $5–10m per point against $25m+ for residential. Management’s own goal is to still be under-earning by 50bp in 2029. The formula rate on which the whole thesis rests is before the Arizona Supreme Court, and the commission that must approve it is elected — its chairman having just been defeated by an anti-rate-increase challenger, with the Attorney General intervening in opposition and RUCO at 9.0–9.2%. The 4.5 GW of “committed” load carries no disclosed minimum take, no term, no collateral and no exit fee, zero subscription contracts have been signed after nine months, and everything enters rate base with no ring-fencing — so a walk-away is a stranded asset in front of an elected regulator. Meanwhile the stock is at a 98.5th-percentile own-history valuation, 22.8x a guided-down year, 1.82x book at a decade high, with a payout heading to ~80%, an ATM ~70% consumed, authorised shares doubled, five points of covenant headroom, ~50–80bp of FFO/debt cushion, and exactly one open-market insider purchase in five years, for $43,270.

11.4 Where I think consensus is offside

First and most important: the market is treating “load growth” and “earnings growth” as the same variable, and management has explicitly told it they are not. The single most useful sentence in four earnings calls is Cooper’s: 1% of residential growth is worth north of $25m, 1% of extra-high-load-factor growth is worth $5–10m. With 4–6 of the guided 5–7 points coming from large C&I, a 5–7% headline is worth roughly what a 1.5–2% residential-only number would be. This is disclosed, quantified, and largely absent from the narrative that has driven a 21.8% year-to-date re-rating.

Second: the FRAM is being priced as a decision when it is currently a legal question — and APS’s own witness says so. The Court of Appeals remand in November 2025 means pending cases may not be able to use the expedited process until legality is settled, and the Arizona Supreme Court has not ruled. Investors underwriting “5–7% becomes evergreen” are underwriting an appellate outcome most have not read. The most striking corroboration comes from APS’s own cost-of-capital witness, Chris Bauer, in sworn direct testimony in the docket: “some equity investors have already begun to evaluate APS based on an assumption that the Company will operate with a formula rate in the coming years and purchasing stock on that basis.” The company’s own expert is telling the Commission that the market has already paid for a mechanism that is unapproved, opposed on design by Staff, conditioned by Staff on a flat ROE, and subject to an unresolved challenge to the ACC’s authority to have created it.

Two further points on the FRAM that the bull case tends to skip. First, ACC Staff’s own footnote makes the trade explicit: “if the FRAM is adopted, the Commission should consider an ROE at the lower end of Staff’s recommended range (9.55%–9.80%).” The regulator has put a swap on the table — you may have the mechanism that closes the lag, but only if your authorised return stays where it is. Second, the ACC’s formula-rate policy requires the mechanism to be built on a historical test year and to include an earnings test holding the utility within ±20bp of authorised, with re-litigation in a full rate case at least every five years. Combined with the LFCR provision that sets recovery to zero if the utility over-earns, the construct is engineered so that operational outperformance is refunded rather than retained. Even in the bull case, the FRAM does not let APS earn above its authorised return; it only stops it earning below. Upside is capped by rule; downside — disallowances, prudence review, wildfire — is not.

Bauer’s testimony also corroborates the earning gap from the company’s side, and at a wider figure than the consolidated numbers suggest: “APS has earned approximately 200 basis points below its authorized ROE.” The reconciliation matters — ~200bp is the ACC-jurisdictional utility gap, against the ~177bp measured at APS GAAP level and ~85bp at PNW consolidated. The difference is the 10.75% FERC transmission book and holdco leverage. Strip both out and the ACC-regulated utility — roughly 79% of rate base — earns something close to 7.5–7.6% against a 9.55% authorisation.

Third: the elected-commission variable is genuinely un-priced. Comparable published utility analyses almost never treat this variable; Southern Company, before Georgia’s elected PSC, is the rare exception. Arizona’s chairman lost his primary four days before this report, to a candidate campaigning on rates, and votes a 14.75% increase in December. The market has no framework for this because the peer set is almost entirely appointed.

Fourth: the historical asymmetry. The 2021 adverse order cost 8% in a single session. The 2024 favourable order — a $491.7m increase and an ROE restored to 9.55% — was met with a ~4.7% decline over the following week. If good Arizona orders have historically been worth nothing and bad ones worth −8%, the distribution around December 2026 is unattractive from a 98.5th-percentile starting valuation.

Where consensus may be offside in the bulls’ favour, and I hold this genuinely: H.B. 2201 is a large, under-discussed structural improvement. Analyses of PG&E and Edison International identify California’s inverse-condemnation regime as the dominant risk in western utilities. Arizona has legislated close to the opposite, with a compliance-based safe harbour and a bar on inverse-condemnation claims, plus securitisation. For a hot, dry, wildland-urban-interface state that is a meaningful reduction in the left tail, and it barely features in the discussion. The FERC transmission mix shift is similarly under-modelled.

11.5 The factor and positioning read

The quantitative positioning evidence supports the framing rather than contradicting it. In the All-Factors model (R² 0.713) PNW loads Sector: Utilities +0.753, Market +0.585, LowVolatility +0.567, Growth −0.535, “Regulated Utility Giants” +0.329, Quality −0.316, with Value zeroed out entirely. Roughly 71% of return variance is factor-explained; five of its twenty nearest factor neighbours are utility ETFs. This is a duration-and-defensiveness vehicle, not a story stock.

The Momentum question resolves in an instructive way. PNW’s Momentum loading is +0.241 in the base model but decays to +0.101 with sectors, +0.073 with industry and +0.038 in All Factors. Under sequential orthogonalisation that decay means the Utilities sector factor and the Regulated Utility Giants basket absorb essentially all of it: PNW has not earned its momentum, it has borrowed it from the basket. The right framing is therefore not “crowded single-stock momentum” but “crowded sector positioning with no company-specific alpha to defend it” — a harder claim, because the exit risk is a regime rotation PNW cannot control rather than a company stumble it might avoid.

The regime read sharpens it. The Regulated Utility Giants basket is the single most extended factor in PNW’s exposure set: +4.99% / z +2.27 (21d), +7.63% / z +2.05 (126d), +12.39% / z +2.84 (252d) — three windows at or beyond the model’s extreme threshold. Notably the generic Utilities sector factor is not extended (252d −4.06%, z −0.31): the crowding is concentrated specifically in large-cap regulated names, exactly where PNW sits. One genuine cross-current: the InterestRate factor ran +5.12% (z +1.91) over 21 days against PNW’s −0.177 loading — PNW has advanced into a duration headwind, meaning the last leg was paid for by the load-growth narrative rather than by falling rates.

Finally, the honest horizon comparison: y1 return +20.0% at a 1.07 Sharpe with only an −8.4% maximum drawdown, against a y5 Sharpe of 0.36 (−26.7% drawdown), a y10 Sharpe of 0.22 on a +7.1% annualised return (−39.3% drawdown), and a lifetime maximum drawdown of −49.5%. The last twelve months sit roughly five times better on risk-adjusted terms than the decade this business has actually delivered. That is a regime artefact, not evidence of a changed business — and it is the cleanest statement of why the current multiple is fragile. (Third-party statistical estimates; loadings and returns are facts, the persistence read is interpretation and regime-dependent.)

11.6 The 3–5 assumptions that matter most

1. The ACC’s December 2026 order on ROE and revenue requirement.
2. Whether a Formula Rate Adjustment Mechanism is granted — and survives the Arizona Supreme Court.
3. Whether large-load growth converts into rate base on terms that protect the utility (minimum take, term, collateral) or leaves stranded-cost risk with ratepayers and therefore with the regulator.
4. Whether the earned-versus-authorised ROE gap actually narrows, versus management’s own 2029/50bp aspiration.
5. Whether the sector’s record valuation holds — because PNW’s funding model depends on issuing equity at a premium to book.

12. Fact vs. Interpretation

13. Open Questions

1. What are the actual commercial terms of the 4.5 GW of committed load? Minimum-take percentage, contract term, collateral, exit fees and credit support are disclosed nowhere — not in the 10-K, not on four earnings calls. The most likely source is ACC Docket E-01345A-25-0105 rate-design testimony. This is the single most important missing datum in the bull case.
2. Will the ACC vote before or after the November 2026 general election? Company guidance indicated October 2026; press reporting from the hearing indicates a November ROO and a December vote. The distinction is materially outcome-relevant and unresolved.
3. Will the Arizona Supreme Court take the formula-rate case, and how will it rule? The entire “5–7% becomes evergreen” premise depends on it.
4. What is APS’s ACC-jurisdictional regulatory earned ROE, as computed in a rate filing? The 7.78% used here is APS GAAP net income over GAAP equity, which includes FERC and non-jurisdictional equity. The direction is corroborated independently; the precise number should come from the docket.
5. Has APS’s Comprehensive Wildfire Mitigation Plan been approved by the Arizona DFFM? The H.B. 2201 safe harbour is conditional on approval and compliance; review was “underway” at the FY2025 10-K, with plans due to the State Forester by 1 May 2026. Insurers challenged plan adequacy in March 2026.
6. Will the IRS accept APS’s revenue-requirement methodology for nuclear PTC “gross receipts”? $73.0m of 2024–25 cash-tax benefit is fully reserved.
7. What is the recovery treatment and amortisation period for the $151.5m regulatory asset created by the Palo Verde sale-leaseback buy-in? No period is stated and recovery is not yet approved.
8. How large will the next equity authorisation be? Only ~$270m of the $900m ATM remained at 5 June 2026 against a $7.95bn three-year plan, and authorised shares were doubled to 300m.
9. What are the rating agencies’ actual adjusted FFO/debt figures and thresholds? Management cites a Moody’s downgrade threshold of 14% and describes itself as “high 14s”; the agencies’ own adjusted calculations are not in the filings.
10. Does the 2026 IRP (due August 2026) include only committed load, and what does it imply for capex? Management has indicated it will include only committed volumes.
11. Will the IBEW agreement (amendable 1 April 2026, with a bargaining notice served) settle without a material cost step-up?
12. How much of the ~20 GW queue is duplicative? Management itself says this is “to be determined.”

14. What Must Be True

14.1 For the bull case

14.2 For the bear case

14.3 The single test that resolves most of it

The Arizona Corporation Commission’s vote on docket E-01345A-25-0105, expected December 2026. It simultaneously determines the ROE, the revenue requirement, whether a formula rate exists, how large loads are priced, and whether the LFCR survives — and it does so before a five-member elected body in the month after a general election in which one incumbent has already been defeated on an anti-rate-increase platform. Almost every disagreement between the bull and bear cases collapses into that one decision, and the parties’ filed positions bracket it widely: APS at $694m and 10.70%, Staff at $506m and 9.55–9.80%, RUCO at $200–278m and 9.00–9.20%, and the Attorney General intervening in opposition.

15. Source Appendix

The full source list with URLs and access dates appears in Appendix B at the end of this article. Primary sources are the Pinnacle West/APS FY2021–FY2025 Forms 10-K and the Q1-2026 Form 10-Q; the DEF 14A proxies of 2024, 2025 and 2026; the full 8-K corpus from July 2021 to July 2026 including the May-2026 investor presentation; all 202 Form 4s and 10 Form 144s; the Q2-2025 through Q1-2026 earnings-call transcripts; SEC EDGAR XBRL company facts; Arizona Corporation Commission docket and decision records; Arizona legislative records for H.B. 2201 and H.B. 2679; S&P Global/RRA regulatory benchmarking; and market data from the AZI price and valuation-index feeds and the FactorsToday factor model.

Published 25 July 2026. All figures are reconciled to primary SEC filings where available. Third-party aggregated data is used only as a cross-check and is identified as such; where an aggregator and a filing disagree, the filing governs and the discrepancy is noted. This article is general information and does not constitute investment advice. The author may or may not hold a position in any security mentioned.

APPENDIX A — Standard Diligence Questionnaire

Pinnacle West Capital Corporation (NYSE: PNW) · 25 July 2026

Supplemental to the research memo. Grounded in the underlying research; Fact / Interpretation / Assumption labelled where it matters. Where a question does not map to a regulated utility, the correct sector analog is given.

General

What thoughtful questions have other investors asked about this company?

Three recur, and all three are the right questions.

“Why doesn’t the best load growth in the sector show up in earnings?” — asked directly on the Q1-2026 call, where an analyst put the ~200bp rate-base-to-EPS gap to the CFO. FACT: Cooper did not dispute it, answering “we will have to revisit all of this at the conclusion of the rate case.” The answer, developed in the growth section and the financial-quality section of the memo, is threefold: the growth arrives in the lowest-margin customer class, more than half of headline revenue growth is fuel pass-through, and Arizona’s 30-month lag means the capital serving the growth earns nothing until a case is decided.

“When do you actually earn your allowed return?” — Julien Dumoulin-Smith on the Q1-2026 call. FACT: CEO Geisler’s answer is the most important management statement in the file: the goal is to earn “within that 50 basis points, given there is some element of structural lag that will continue to exist… by 2029 and going forward.”

“Is the formula rate real?” — the question the market is least well informed on. FACT: the ACC’s formula-rate policy statement was remanded by the Arizona Court of Appeals on 21 November 2025 and review is pending at the state Supreme Court. FACT: APS’s own cost-of-capital witness testified that “some equity investors have already begun to evaluate APS based on an assumption that the Company will operate with a formula rate… and purchasing stock on that basis.”

A fourth that is not being asked, and should be: which loads are actually in APS territory rather than SRP’s or TEP’s (the growth section), and what the commercial terms of the 4.5 GW of “committed” load actually are — undisclosed anywhere.

Cyclicality and Earnings Nature

Are earnings at a cyclical high or low? INTERPRETATION: at a regulatory low, not a cyclical one — and that distinction is the investment case. FY2025 diluted EPS of $5.05 sits below FY2021’s $5.47, and FY2026 is guided down to $4.55–$4.75. APS earned a 7.78% GAAP ROE against a 9.55% authorisation. On the company’s own sworn testimony the ACC-jurisdictional utility earns ~200bp below authorised. There is roughly $1.29/share of earnings available from simply closing that gap, requiring no new capital — only a regulatory decision — plus ~$0.62 if the requested ROE step is granted. Earnings are therefore depressed relative to the allowed level, and the recovery is a regulatory event with a date on it (December 2026), not an economic cycle.

Driven by the external environment or internal actions? Overwhelmingly external. Operationally the company has performed: a 92% Palo Verde capacity factor (30m+ MWh for a 17th consecutive year), 93.1% non-nuclear summer availability, top-quartile reliability, customer satisfaction moved from fourth quartile in 2020 to first/second in 2024–25, and core O&M held roughly flat ($955m 2024A → $978m 2025A → $970–980m 2026E) against a 2.2–2.4% growing customer base. The earnings shortfall is regulatory: test-year lag, the fair-value construct, and the 2021 order. INTERPRETATION: management controls the numerator poorly and the denominator not at all.

How stable are revenues? Very. Retail electricity is essential, non-discretionary and non-storable, sold into a 100%-share franchise. FY2025 bad debt was $28.6m on $5,339.9m of revenue — 0.54%. The main volatility is weather: FY2025 carried a $0.71 year-over-year weather drag, and Q1-2026’s $0.27 (versus a $(0.04) loss) included $0.13 of weather on the hottest March on record. Fuel volatility is passed through via the PSA, capped at ±$0.006/kWh annually.

Outlook for products/services? Structurally growing for the first time in two decades. Weather-normalised retail sales have grown 4–7% for nine consecutive quarters; guidance is 4–6% for 2026 and 5–7% through 2030.

How big will this market be — growing, shrinking, domestic or international? Purely domestic and single-state. FY2025 system peak was 8,648 MW (+5.3%); APS discloses 4.5 GW of committed load and ~20 GW of uncommitted opportunity. Governor Hobbs’s April 2026 energy plan contemplates data-centre load nearly tripling APS+SRP demand, requiring up to 29,000 MW. ASSUMPTION: applying a 10–25% realisation rate to the uncommitted queue — calibrated against TEP’s local ~30:1 queue-to-contract ratio — implies roughly 2–5 GW of incremental realised load by the mid-2030s, about one quarter of the headline narrative.

Business Quality and Competitive Moat

Is the industry getting more or less competitive? Less, and in Arizona it is now legally settled. The 2004 Arizona Court of Appeals decision voided the ACC’s retail-competition rules as unconstitutional and voided every competitive provider’s operating authority; the Legislature repealed the 1998 deregulation statute outright in April 2022; and the ACC administratively closed the retail-competition docket on 27 August 2024 for “inactivity and obsolescence.” PNW removed the historical discussion from Item 1 of the FY2025 10-K. Retail competition in Arizona is dead.

How profitable is the business (ROIC, ROE)? Poorly, and this is the central finding. ROIC: 6.66% (2015) → 6.14% (2017) → 5.69% (2020) → 4.31% (2022) → 4.61% (2025) — down ~205bp in a decade. ROE has not exceeded 10.60% in eleven years; the 2021–25 average is 9.00% against 9.78% for 2016–20. FY2025 consolidated ROE was 8.93%; APS itself earned 7.78% against a 9.55% authorisation. Across seventeen regulated utilities reviewed for comparison, ROIC spans ~5.0–6.6%; PNW’s ~4.6% is the joint lowest, matched only by a pure-wires utility on a construct already regarded as punitive. A vertically integrated utility carrying generation risk should sit above the median, not at the wires floor. Caveat, stated honestly: FY2025 ROIC is mechanically depressed by a ~$2.0bn step-up in operating-lease right-of-use assets as renewable and storage PPAs were capitalised. The decade trend is real; the final-year step is partly artefact.

How profitable is the industry — how many competitors, what barriers to entry? The barrier is the most absolute available: government-conferred legal monopoly (Greenwald’s fourth category). Entry is not expensive, it is illegal. But it is also the least valuable barrier, because the authority that erects it sets the return behind it. By Greenwald’s own numeric benchmark — 15–25% ROIC indicates advantages present, 6–8% indicates advantages absent — PNW at 4.6% sits below the band that signals advantages are absent. The rent is real and large; it is captured by ratepayers and the regulator.

Can the business be easily understood? Yes. Earnings = rate base × equity layer × allowed ROE. The complexity is entirely in the regulatory overlay: the fair-value increment, seven named adjustor mechanisms, and a formula-rate proposal whose legality is before the state Supreme Court.

Can it be undermined by foreign low-cost labour? No. Electricity delivery is inherently local and non-tradeable. The relevant analog is behind-the-meter self-supply — named in the 10-K as a competitive risk: “large customers developing large, utility scale generation to serve their energy needs.” That is the genuine ten-year threat to the load-growth thesis.

Do brands matter? No. Customers cannot choose their supplier. Customer satisfaction matters only instrumentally, as regulatory goodwill — and it is a live incentive metric (J.D. Power scores carry 20% of the shared enterprise bonus weighting).

What is the nature of competition? Not for customers but for the regulator’s favour, and for marginal new large load against Salt River Project. SRP is a political subdivision exempt from ACC jurisdiction, serving ~1.06m residential customers in the metro core, with no shareholders, no authorised ROE, no state income tax and tax-exempt debt. On a consistent EIA-861 basis for calendar 2024, its residential rate was 13.46¢/kWh against APS’s 16.45¢ — SRP is 18.2% cheaper (20.2% across all classes). Note the bill gap is much narrower (−4.1%) because SRP customers use ~17% more kWh. SRP voted itself a 2.4% increase while APS litigates for 14.69%. INTERPRETATION: for the marginal incremental large load — which is where all the growth is — APS does not have pricing power. SRP is the ceiling.

Customers’ switching costs? Infinite for existing retail customers (they cannot switch). Effectively zero for the incremental large load that drives the entire growth thesis — a hyperscaler chooses its site, and with it its utility, before construction.

Financial Condition and Balance Sheet

Assets not fully recognised on the balance sheet? Three. (i) The operated 29.1% Palo Verde interest (1,146 MW) carries at depreciated original cost, far below replacement value; APS filed a Notice of Intent with the NRC in March 2026 for subsequent licence renewal extending the units to roughly 2065–2067, which if granted adds two decades of rate-base annuity not in any current book value. (ii) $73.0m of nuclear production tax credit claimed on the 2024 and 2025 returns is fully reserved as an uncertain tax position with no earnings benefit recognised — unrecognised tax benefits rose from $44.2m to $105.5m in 2025. (iii) The franchise itself carries no balance-sheet value.

Off-balance-sheet liabilities? Materially fewer than before ASC 842, but the disclosure that matters is forward. FACT: $11.4bn of leases are executed but not yet commenced, primarily energy storage, commencing April 2026 to June 2028 — a figure approaching the entire current rate base, already landing at +$410.6m in Q1-2026. Nine storage PPA operating leases (+$2,195.7m of ROU assets) commenced in 2025 on 20-year terms, stated in the 10-K as recoverable through the PSA. Also: Palo Verde decommissioning obligations (trust assets $1,414.2m), coal reclamation, and a $7.3bn/25-year anchor-shipper commitment on the Transwestern pipeline. A revealing detail: on 18 February 2026 both credit agreements were amended to exclude PSA-recoverable PPAs and storage leases from the covenant definition of indebtedness. Management negotiated a carve-out for exactly the obligation that is exploding; banks granted it, and rating agencies generally do not grant the equivalent.

How conservative is the accounting? Mixed, with two specific soft spots. Conservative: no pension contributions required 2026–28 under an LDI strategy; the nuclear PTC benefit fully reserved rather than recognised; the Clear Creek wind investment written to zero in Q4 2022. Aggressive or at least flattering: AFUDC was $108.9m in FY2025, or 17.7% of net income to common — roughly one dollar in six of “earnings” is a non-cash capitalised construction return that becomes cash only when plant enters service and the ACC puts it in rates, and the $22m year-on-year rise in AFUDC-equity more than accounts for the entire $8m earnings improvement. With CWIP at $1,649.5m this share rises. Second, a $151.5m regulatory asset was created by the Palo Verde sale-leaseback buy-in with no stated amortisation period and recovery not yet approved. Total regulatory assets of $1,749m are described in PNW’s own critical-accounting language as reflecting “the current political and regulatory climate in Arizona and is subject to change” — the company’s own word.

How CapEx-hungry is the business? Extremely. $7.95bn of APS capex over 2026–28 ($2,600m / $2,650m / $2,700m) against FY2025 operating cash flow of $1,805.1m. FY2025 capex of $2,624.6m equalled ~21% of the $12.5bn rate base in a single year. Operating cash flow covered 68.8% of FY2025 gross capex and 67.9% over five years. This is the defining financial characteristic of the business.

Capital Allocation and Management

How much FCF does the business generate, how does management use it, what is the philosophy? It generates none. Post-capex free cash flow was negative in all five years 2021–2025: −$507.8m, −$328.7m, −$457.8m, −$328.0m, −$513.1m, cumulatively −$2,135.4m. After $1,952.4m of dividends the five-year external funding requirement was −$4,087.8m, closed with $3,510.1m of net new long-term debt, short-term borrowings and ~$0.5bn of equity — roughly 92% debt. The philosophy is the standard regulated-utility one: raise external capital, put it in rate base, earn the allowed return. It works only where the allowed return is actually earned.

Significant acquisitions recently? None of consequence, and the record of trying is poor. Bright Canyon Energy was sold in 2024 (~$84.3m of proceeds, a $22,988k gain); the 9.9% Clear Creek wind investment was fully impaired in Q4 2022. The most significant recent transaction is a buy-in: ~$198.7m in September 2025 for two of three Palo Verde Unit 2 sale-leaseback interests (~94 MW), reducing non-controlling interests from $103.2m to $40.6m. INTERPRETATION: on the merits this is the clearest example of good capital allocation in the file — repurchasing a slice of the single best asset in the fleet at a moment when firm carbon-free capacity is scarce. The caveat is that recovery of the resulting $151.5m regulatory asset is not yet approved.

Buying back shares? No. There is no repurchase programme, no issuer-purchases table in Item 5, and no authorisation anywhere in the five-year 8-K corpus. PNW is a structural net issuer.

Issuing large amounts of new shares to insiders? Not to insiders — but issuing heavily to the market. Average diluted shares went 113.2m (2021) → 122.0m (2025), +7.8%. In February 2024 PNW sold forward 11,240,601 shares — ~9.9% of the company — at $64.505, roughly 1.18x then-book and near the cycle low; 4,426,350 remain undelivered and settlement has been extended twice. A $900m ATM opened in November 2024 was ~70% consumed by June 2026, with ~$264m of new forwards struck in Q1-2026 alone at $98–101. A $525m 4.75% convertible due June 2027 sits ~15% in the money at a $92.30 conversion price. Authorised shares were doubled from 150m to 300m in May 2025. To management’s credit, the 2026 forwards at $98–101 are struck far better than the 2024 block.

Compensation policy of directors/management? This is where the behaviour is explained, and it is the sharpest governance finding in the file. The annual incentive is 50% earnings-dollars against a committee-set target plus 50% business-unit performance — the latter including “PNW CAPITAL,” a capital-spend-versus-budget metric, which paid 200%. The LTIP is 40% relative TSR versus the EEI Index, 40% EPS against an internally set target, and 20% “MW Installed.”

There is no earned-ROE metric, no authorised-return-achievement metric, no ROIC metric and no regulatory-lag metric anywhere in either plan. Every financial metric is satisfied by growing the asset base and spending the plan; none requires earning the authorised return on it. That is a mechanical explanation for assets +36.5% against flat earnings.

The most damning specific, verbatim from the 2026 proxy: “For 2025, we set the APS earnings target at $622 million, a decrease to 2024 actual incentive earnings of $641.9 million. Likewise, we set Pinnacle West’s 2025 earnings target at $544 million, a decrease to Pinnacle West’s 2024 actual incentive earnings of $608.8 million.” Actual PNW incentive earnings came in at $594.7m — still below 2024 — and paid at 182% of target. On the 10-K’s own performance graph, $100 invested at end-2020 was worth $138 (PNW) versus $144 (EEI) and $195 (S&P 500); yet the 2022 performance-share grant paid 186.5% and the 2023 grant 159%. Say-on-pay support: 94.8%, 94.8%, 96.6%, 97.3% (2026) — no shareholder pressure at all. One recent item: a $1.5m off-plan retention award to the Chief Nuclear Officer on 23 June 2026 with no performance condition.

Motivations of management? Insiders own 0.19% of the company (20 directors and officers, 229,489 shares, ~$24.3m). Over five years, parsing all 202 Form 4s: exactly one open-market purchase — director Susan Flanagan, 500 shares at $86.54, $43,270 — against 19 sales totalling 47,283 shares and $4.09m. Dollars sold to bought is 72:1. Not one Form 4 references a 10b5-1 plan; all 129 filings carrying the checkbox report none, so every sale was discretionary. The COO sold 24,809 shares and holds 5,160, including 6,567 at $99 on 13 May 2026, mid-hearing. INTERPRETATION: management has essentially no personal economic exposure to the capital plan it is executing, and is paid on megawatts installed and capital spent rather than on returns earned. It is doing exactly what it is paid to do.

Valuation and Market Data

Is the stock an ADR, MLP, or K-1 issuer? No. Pinnacle West Capital Corporation is a US domestic C-corporation, NYSE-listed, with a single class of common stock and standard Form 1099 dividend reporting. No K-1, no ADR, no MLP structure, no dual-class shares, no preferred stock outstanding.

Dividend policy? Indicated annual dividend ~$3.70 (Q2-2026 declared at $0.91), a ~3.5% yield, raised for a 14th consecutive year. The rate has risen by exactly $0.06 every year for five years, producing mechanically decelerating growth: +2.41%, +1.76%, +1.73%, +1.70%, +1.68%. Payout ratios: 59.7% (2021), 78.3%, 77.1%, 64.8%, 68.6% (2025); on FY2026 guidance the payout rises to roughly 80%. FACT: no numeric payout target appears anywhere in the 10-K — the “sustained 65–75%” figure is an investor-deck construct, not a filed commitment, and 2026 guidance already breaches it. FACT: 100% of the five-year dividend was externally funded — the yield is a return of capital raised from the markets, not a distribution of cash the business generated.

How profitable is the business? Answered above: ROE 8.93% consolidated (APS 7.78%), ROIC 4.61%, net margin 11.5%, operating margin 20.0% (31.3% normalised for fuel pass-through). Against an authorised 9.55% and a requested 7.63% WACC, the business does not demonstrably earn its cost of capital.

Is net income diverging from cash from operations? Not adversely — the ratio is healthy and typical for a utility. FY2025 CFO of $1,805.1m against net income of $631.6m is a 2.86x ratio, driven by $969.6m of D&A. The divergence that matters is the other one: CFO against capex. Operating cash flow covered only 68.8% of FY2025 gross capex. One tailwind is now spent: the deferred fuel regulatory asset unwound from $463.2m (2023) to $149.1m (2025), collecting $463.0m from customers in 2025 — support to 2023–25 operating cash flow that does not repeat.

Risks and Downside

What factors would cause the stock to decline? In order of probability-weighted impact: (1) an adverse December 2026 ACC order — an ROE at or below 9.55% with the FRAM denied or deferred, against Staff at $506.5m/9.55–9.80%, RUCO at $200–278m/9.00–9.20% and the Attorney General intervening in opposition; (2) an Arizona Supreme Court ruling voiding the formula-rate policy; (3) multiple compression from a 98.5th-percentile own-history valuation, which is dangerous here because the equity-funding model itself depends on issuing at a premium to book; (4) a credit downgrade — the FFO/debt cushion is ~50–80bp against Moody’s 14% threshold, below management’s own 100bp aspiration; (5) large-load disappointment or a stranded-cost event; (6) rising long rates, given a −0.177 interest-rate factor loading.

Risk of a catastrophic loss? Low but not zero, and it is concentrated in one asset. A prolonged Palo Verde outage would be severe: 1,146 MW of entitlement, 22.7% of delivered energy, operated by APS. Mitigants are strong — a 92% capacity factor, 30m+ MWh for a 17th consecutive year, and a 2025 INPO award.

Wildfire is the classic western-utility catastrophic risk, and Arizona has legislated it down. H.B. 2201, signed May 2025 and effective September 2025, bars inverse-condemnation claims against electric utilities, bars punitive and consequential damages, and creates a mitigation-plan safe harbour limiting liability absent clear and convincing evidence of willful, intentional or reckless misconduct; H.B. 2679 authorises securitisation. This is close to the structural opposite of the California regime that bankrupted PG&E, and it is materially under-discussed. Caveats: the safe harbour requires plan approval and compliance, it is untested in Arizona courts, insurers challenged plan adequacy in March 2026, and the ACC separately denied APS’s wildfire cost deferral on 18 June 2025.

Chance of a total loss? Very low. A rate-regulated monopoly with 1.43m captive customers, an essential product, 0.54% bad debt, investment-grade ratings at both entities (APS Baa1/BBB+/BBB+, PNW Baa2/BBB+/BBB, all Stable and reaffirmed April 2026), five points of covenant headroom against a 65% debt/cap limit, and $868.0m of undrawn revolver capacity at year-end 2025 rising after the February 2026 upsizing. The realistic bad outcome is a de-rating and years of sub-cost-of-capital returns, which is what 2021–2023 delivered — not impairment of the enterprise.

Recent News and Events

Has the business environment changed recently? Yes, in both directions, and the balance has deteriorated over the last eight months. Improving: nine consecutive quarters of 4–7% weather-normalised sales growth with Q1-2026 at 9.4% and C&I at 14.6%; three consecutive record system peaks; H.B. 2201’s wildfire safe harbour and H.B. 2679’s securitisation authority (both 2025); a UNS Gas formula rate granted in February 2026 at ~86% of ask; and credit outlooks reaffirmed by all three agencies in April 2026. Deteriorating: the Arizona Court of Appeals remanded the formula-rate policy on 21 November 2025, putting the entire lag-reduction thesis sub judice; the ACC denied the wildfire deferral, capped APS equity infusions at 2.5% of assets, and cut the DSM budget to $40m; FY2026 EPS is guided down to $4.65; and on 21 July 2026 ACC Chairman Nick Myers was defeated in his own party’s primary by an anti-rate-increase challenger, five weeks before he votes on a 14.69% increase.

Significant acquisitions? None. The relevant transaction is the ~$198.7m Palo Verde sale-leaseback buy-in (September 2025). Bright Canyon was sold in 2024.

Change in accounting policies? No change in policy, but two changes in presentation that matter. Nine energy-storage PPAs were capitalised as operating leases in 2025 (+$2,195.7m of ROU assets) with $11.4bn more executed and not yet commenced — this inflates apparent “debt” and mechanically depresses ROIC and ROA without economic deterioration, since the costs are PSA-recoverable. And accumulated depreciation fell from $9,412.2m to $9,234.0m despite $969.6m of D&A — this is the Cholla retirement (~$750m of gross plant charged to accumulated depreciation under the standard convention), not an impairment or a write-off; only $81m of net book value remained and it continues to earn a return in rates.

Recent changes — new markets, facilities, management? Management: CEO succession announced 11 December 2024 — Jeff Guldner to Ted Geisler, effective 1 April 2025. Andrew Cooper remains CFO. The IBEW agreement became amendable on 1 April 2026 with a bargaining notice served. Facilities: Cholla Units 1 and 3 retired 30 April 2025; +90 MW at Sundance in service 2025; +397 MW at Redhawk under construction for 2028; up to 2,000 MW of flexible gas planned (Desert Sun at Gila Bend, Phase 1 late 2030); anchor-shipper commitment on the Transwestern Desert Southwest pipeline for 2029/30; a November 2025 all-source RFP seeking ≥1,000 MW for 2029–31; and a March 2026 NRC Notice of Intent for Palo Verde subsequent licence renewal to ~2065–2067. Markets: no new geographies — PNW remains single-state. The genuinely new market is the large-load subscription product (Tranche 1, 1.0–1.2 GW), on which zero contracts had been signed as of the 4 May 2026 call after roughly nine months of negotiation. Rules: the ACC voted 5-0 on 4 March 2026 to repeal Arizona’s Renewable Energy Standard (Decision No. 81677) and 4-0 on 8 July 2026 to repeal the Electric Energy Efficiency Standards. The renewable repeal is NOT in effect — the Attorney General disapproved the final rulemaking package under A.R.S. the relevant section-1044 in late May 2026 and the ACC has authorised litigation; PNW’s Q1-2026 10-Q says “APS cannot predict the outcome of this matter.” The earnings impact either way is close to nil — the RES surcharge and DSMAC are pass-through adjustors, so removing them removes matching revenue and cost, and the CEO called the renewable repeal “no impact” on the Q1-2026 call. What it does create is customer-bill headroom: roughly $110m/yr of RES surcharge plus a DSM budget cut from ~$91m to $40m is on the order of $160m of annual bill relief — genuinely useful in the middle of a 14.75% rate request, and arguably worth more to PNW than the adjustors were. The residual risk is that repealing the mandate does not repeal the above-market solar PPAs signed to comply with it, and that losing the DSMAC removes a recovery mechanism the 2022 rate case had only just created. Strategy: on 7 August 2025 APS abandoned its zero-coal-by-2031 target, moving the Four Corners exit to “no later than 2038,” and replaced “zero-carbon by 2050” with “carbon-neutral by 2050,” dropping interim targets.

APPENDIX B — Source Appendix

Pinnacle West Capital Corporation (NYSE: PNW) · 25 July 2026

All sources accessed 25 July 2026 unless otherwise stated. Primary SEC filings govern; third-party aggregated data is identified as such and used as cross-check only. Where an aggregator and a filing disagree, the filing governs and the discrepancy is noted in the memo.

1. SEC filings — primary (mirrored locally to output/PNW/sources/)

The trailing 60-month corpus was enumerated via scripts/edgar.sh since PNW 2021-07-01 (431 filings) and mirrored via scripts/fetch_sources.sh: 5× 10-K, 15× 10-Q, 101× 8-K, 5× DEF 14A, 9× DEFA14A, 202× Form 4, 10× Form 3, 10× Form 144, plus 11-K, S-3ASR, S-8, 424B5 and ARS. CIK 0000764622.

Annual reports (Form 10-K), Pinnacle West Capital Corp / Arizona Public Service Co combined:

• FY2025, filed 2026-02-25, accession 0000764622-26-000011 — pnw-20251231.htm — the principal source for this memo. Item 1 Business pp.3–31; Item 1A Risk Factors pp.32–46; Item 2 Properties pp.47–50; Item 5 (performance graph); MD&A pp.55–80 (Regulatory Overview pp.61–63; Operating Results pp.65–70; Liquidity and Capital Resources pp.70–75, including the capital-expenditure table p.72, equity forwards p.73, covenants p.74 and credit ratings p.75); Note 1 Property, Plant and Equipment and AFUDC pp.110–111; Note 4 Revenue p.119; Note 5 Income Taxes pp.121–124; Note 6 Lines of Credit pp.127–129; Note 7 Long-Term Debt pp.129–132; Note 8 Regulatory Matters pp.132–150; Note 9 Retirement Plans p.148; Note 12; Note 16 Common Stock Equity and EPS pp.175–178; Note 20 Leases pp.189–191. https://www.sec.gov/Archives/edgar/data/764622/000076462226000011/pnw-20251231.htm
• FY2024, filed 2025-02-25, accession 0000764622-25-000023 — pnw-20241231.htm
• FY2023, filed 2024-02-27, accession 0000764622-24-000016 — pnw-20231231.htm
• FY2022, filed 2023-02-27, accession 0000764622-23-000023 — pnw-20221231.htm
• FY2021, filed 2022-02-25, accession 0000764622-22-000014 — pnw-20211231.htm (2019 rate-case outcome and prior adjustor terms)

Quarterly reports (Form 10-Q):

• Q1-2026, filed 2026-05-04, accession 0000764622-26-000025 — pnw-20260331.htm. Note 7 Regulatory Matters carries the March-2026 intervenor testimony (ACC Staff and RUCO positions in the 2025 Rate Case).
• Q3-2025 (2025-11-03), Q2-2025 (2025-08-06), Q1-2025 (2025-05-01), and the 2021–2024 series.

Proxy statements (DEF 14A) — compensation, incentive metrics, insider ownership:

• Filed 2026-04-03 (FY2025 pay) — annual incentive design p.62, earnings-target reset pp.64–65, shared enterprise metrics p.66, LTIP design pp.68–70, beneficial ownership p.105
• Filed 2025-04-08 (FY2024 pay); filed 2024-04-04 (FY2023 pay)
• DEFA14A a2024defa14abryanretirement.htm, 2024-05-06 — note: this is a Report Feedback Statement rebutting a Glass Lewis “overboarded” recommendation against director Glynis A. Bryan, not an executive retirement announcement

Current reports (Form 8-K) — key items used:

• 2026-05-06, accession 0000764622-26-000029, Item 7.01 Reg FD, Exhibit 99.1 “Powering Arizona’s Future — May Investor Meetings” (mayinvestordeck.htm) — the single richest source for current guidance: long-term EPS growth 5–7% off the original 2024 midpoint; rate-base guidance ACC $12.23bn (2024) → $15.7bn (2028) and FERC $2.52bn → $4.0bn; the 2025 Rate Case litigated-positions matrix; the 2026–28 financing plan; O&M guidance; credit objectives and ratings; committed/uncommitted load. Signed by Andrew Cooper, CFO. https://www.sec.gov/Archives/edgar/data/764622/000076462226000029/mayinvestordeck.htm
• 2026-06-08 (June investor handouts, Ex-99.1); 2026-04-15 (April handouts)
• 2026-02-25 — Q4/FY2025 earnings release (Ex-99.1) and slide presentation (Ex-99.2), including FY2026 guidance of $4.55–$4.75
• 2026-05-19 — Item 5.07, 2026 annual meeting voting results (say-on-pay 97.3%)
• 2026-06-29 — Item 5.02, $1.5m Discretionary Credit Award to the Chief Nuclear Officer
• 2026-06-05 (×2) — ATM equity distribution agreement amendment; 2024-11-08 — ATM programme establishment
• 2025-09-02 — extension of the February 2024 forward sale agreements
• 2025-06-13 — 2025 Rate Case filing; 2024-03-04 — February 2024 forward sale agreements; 2024-06-04/06-06 — convertible notes
• Full 8-K index: output/PNW/sources/filing_index_PNW.txt

Insider filings: all 202 Forms 4/4-A (2021-07 to 2026-05) and 10 Forms 144, parsed from raw XML (transactionCode, transactionShares, transactionPricePerShare, transactionAcquiredDisposedCode, aff10b5One), 1,141 individual transaction lines.

SEC EDGAR XBRL company facts, CIK 0000764622 — Revenues, OperatingIncomeLoss, EarningsPerShareDiluted, PropertyPlantAndEquipmentNet, StockholdersEquity, CostOfGoodsSoldElectric — 10-K facts 2012–2025, via scripts/edgar.sh.

2. Earnings call transcripts — primary

Retrieved via the ROIC.ai MCP (list_earnings_calls, get_earnings_call_transcript); local copies and extracted quotes at output/PNW/transcripts/PNW_transcript_key_quotes_2025Q2-2026Q1.md.

• Q1-2026, 4 May 2026 — Theodore (“Ted”) Geisler (Chairman, President & CEO) and Andrew Cooper (SVP & CFO). Source of the “within that 50 basis points… by 2029” exchange with Julien Dumoulin-Smith; the “duplicative projects or interest” comment on the ~20 GW queue; confirmation that 2026 equity needs are complete; and the ~$850m of priced equity forwards.
• Q4-2025, 25 February 2026 — source of the “1% of residential growth is somewhere north of $25 million, whereas 1% related to extra high load factor could be more in the $5 million to $10 million range” quotation; the UNS Gas formula-rate read-through; the “Moody’s… downward threshold is 14%… we’re high 14s” exchange; and the “substantial regulatory lag in the jurisdiction” comment.
• Q3-2025, 3 November 2025 — “all of our investment goes in the rate base. The subscription model still goes into rate base.”
• Q2-2025, 6 August 2025 — the 2025 Rate Case parameters as filed and the “day 1 of new rates in effect still means meaningful regulatory lag” comment.

Company press release, “Pinnacle West Reports 2026 First-Quarter Financial Results” — https://www.pinnaclewest.com/newsroom/company-news/news-release-details/2026/Pinnacle-West-Reports-2026-First-Quarter-Financial-Results/default.aspx

3. Arizona regulatory and legislative sources

Arizona Corporation Commission — decisions and dockets:

• Decision No. 78317 (2019 Rate Case, decided 2 November 2021): ROE 8.70% including an explicit 20bp penalty; $215.5m Four Corners SCR disallowance; total annual revenue change of −$4.8m. Subsequently reversed in part by the Arizona Court of Appeals; joint resolution 14 June 2023, ACC approval 21 June 2023, restoring the $215.5m, reversing the penalty (8.70% → 8.90%) and creating the Court Resolution Surcharge for $59.6m of lost revenue.
• Decision No. 79293 (2022 Rate Case, approved 22 February 2024, final order 5 March 2024, rates effective 8 March 2024): +$491.7m base revenue; ROE 9.55%; 0.25% return on the fair-value increment; effective fair-value rate of return 4.39%; SRB mechanism approved.
• Decision No. 76295 (2016 Rate Case, Docket E-01345A-16-0036; settlement signed 27 March 2017, ACC approved 15 August 2017 by a 4-1 vote, order 18 August 2017, rates effective 19 August 2017): net retail base rate increase of $94.6m (base rate revenue +$363m before adjustor transfers), authorised ROE 10.0%, capital structure 55.8% common equity / 44.2% debt, typical residential bill impact +4.54%.
• The 2017-2021 causal chain — A.R.S. the relevant section-246 customer complaint filed January 2018 alleging the actual bill impact exceeded the advertised 4.54%; ACC opened a rate review (Docket E-01345A-19-0003) on 9 January 2019 and retained Overland Consulting (~$250,000) to audit APS’s books, a report subsequently withheld on privilege grounds; ACC order of 11 June 2019 required APS to file a rate case by 31 October 2019 on a June 2019 test year and to remediate customer communications. The 20-basis-point ROE penalty in Decision No. 78317 was imposed for customer-service failures — a faulty online rate-comparison tool (Arizona Attorney General consent decree) and below-average call-centre performance — and was vacated on appeal.
• Summer disconnection moratorium — following the September 2018 heat-related death of Stephanie Pullman, disconnected over roughly $51 of arrears, APS voluntarily suspended residential disconnections on 13 June 2019; the ACC enacted emergency rules on 20 June 2019 and final rules on 2 November 2021. Cost to APS: ~$10m pre-tax in 2019 and ~$23m pre-tax in 2020 including COVID-era effects.
• Credit rating actions around the 2019 rate case (PNW FY2021 10-K): Fitch downgraded PNW and APS on 12 October 2021; S&P downgraded on 9 November 2021; Moody’s, having placed both on review 12 October, downgraded on 17 November 2021. Pinnacle West went from Moody’s A3 / S&P A− / Fitch A− to Baa2 / BBB+ / BBB and has not recovered the downgrades; outlooks returned to Stable within five weeks of the March 2024 order (S&P 7 March, Moody’s 20 March with a further PNW notch, Fitch 26 March with a further PNW notch attributed to holdco leverage and structural subordination).
• ACC Decision No. 79293 (2022 Rate Case, Docket E-01345A-22-0144) full text: https://docket.images.azcc.gov/0000210704.pdf — authorised original-cost rate base $10,355,411k, fair value rate base $16,424,540k, fair value increment $6,069,129k, common equity 51.93% at 9.55%, long-term debt 48.07% at 3.85%, WACC 6.81%, FVROR 4.39%, base rate revenue increase $491,678k (+14.56%), net revenue increase $253.4m, base fuel rate $0.038321/kWh.
• Docket E-01345A-25-0105 — the pending 2025 Rate Case. Filed 13 June 2025; hearing opened 18 May 2026 and ran to early July 2026; ALJ recommended opinion expected November 2026; Commission vote expected December 2026. Includes the direct testimony of APS cost-of-capital witness Chris Bauer (“APS has earned approximately 200 basis points below its authorized ROE”; “some equity investors have already begun to evaluate APS based on an assumption that the Company will operate with a formula rate”). Litigated positions are tabulated in the May- and June-2026 investor decks.
• Docket E-00000A-25-0069 — ACC Inquiry into Data Center Rate Classifications; Large Load Users Development Workshop held 16 April 2026.
• ACC Formula Rate Policy Statement, adopted December 2024.
• Decision No. 81677 (4 March 2026) — ACC votes 5-0 to repeal the Renewable Energy Standard and Tariff rules, Docket RE-00000A-24-0026. Sierra Club and the Arizona Attorney General filed rehearing applications 30 March 2026; the Notice of Final Rulemaking went to the Attorney General 2 April 2026 and was disapproved under A.R.S. the relevant section-1044; ACC news release, “ACC to Explore Options After Disapproval of Renewable Energy Standard and Tariff Rulemaking Package,” 1 June 2026, https://www.azcc.gov/news/home/2026/06/01/acc-to-explore-options-after-disapproval-of-renewable-energy-standard-and-tariff-rulemaking-package. The repeal is not in effect. Primary confirmation: PNW/APS Form 10-Q for Q1-2026, “Energy Modernization Plan.”
• ACC final vote to repeal the Electric Energy Efficiency Standards (A.A.C. Title 14, Ch. 2, Art. 24), 8 July 2026, 4-0, Docket RE-00000A-24-0025 — https://www.azcc.gov/news/home/2026/07/08/acc-repeals-outdated-electric-energy-efficiency-standards-rules--eliminating-surcharges
• Phelps Dodge Corp. v. Arizona Electric Power Cooperative, Inc., 207 Ariz. 95, 83 P.3d 573 (App. 2004), No. 1 CA-CV 01-0068 — holding A.A.C. R14-2-1611(A) facially unconstitutional under Ariz. Const. art. 15 §the relevant section and 14, invalidating R14-2-1609©-(J) and -1615(A),© as ultra vires, and invalidating seven further rules for failure to obtain Attorney General certification under A.R.S. the relevant section-1044(B) — the same provision now blocking the 2026 renewable repeal. https://coa1.azcourts.gov/Portals/1/89/opinionfiles/CV/CV010068.pdf
• Arizona H.B. 2101 (2022), signed 26 April 2022 — repealing the 1998 statutory framework for retail electric generation competition. Retail competition in Arizona is legally foreclosed. Solar Grid Access Charge affirmed on limited rehearing 17 December 2024. Retail-competition docket administratively closed 27 August 2024. RES repeal ROO approved 4 March 2026 (Decision No. 81677). DSM budget reduced to $40m, 3 December 2025.
• ACC newsroom: https://www.azcc.gov/ — including “Maricopa County Judge Dismisses Lawsuit Against ACC Regarding Formula Rate Policy,” 13 June 2025; and the chairman-selection release, 14 January 2026: https://azcc.gov/nick-myers/news/2026/01/14/commissioner-nick-myers-selected-as-chairman-of-the-arizona-corporation-commission

Arizona legislation:

• H.B. 2201 (57th Legislature, 1st Regular Session) — wildfire mitigation plans and utility liability. Senate Fact Sheet: https://www.azleg.gov/legtext/57leg/1R/summary/S.2201FIN_ASPASSEDCOW.DOCX.htm
• H.B. 2679 — securitisation authority, signed 13 May 2025.
• Bill tracking: https://fastdemocracy.com/bill-search/az/57th-1st-regular/bills/AZB00018314/

Arizona courts:

• Arizona Court of Appeals, remand of the formula-rate policy challenge (RUCO/ALCG), 21 November 2025 — petitions for review pending at the Arizona Supreme Court.
• Arizona Court of Appeals Div. 1, Vote Solar / State of Arizona / AriSEIA / SEIA v. Arizona Corporation Commission, Nos. 1 CA-CC 25-0001, -0002, -0003, -0004 (consolidated), memorandum decision 16 June 2026 (Kiley, J.) — VACATING the solar Grid Access Charge on due-process grounds and remanding. ACC motion for reconsideration filed 24 June 2026; APS motion filed 1 July 2026; both pending and no mandate issued as at 25 July 2026, so the ruling is not final. No Pinnacle West 8-K discloses the ruling; first SEC disclosure is expected in the Q2-2026 10-Q.
• Arizona Public Service Company v. Sierra Club / RUCO, No. 1 CA-CC 21-0002 (Ariz. Ct. App. Div. 1, 7 March 2023) — affirming the 8.9% base ROE and the fair-value-increment calculation, vacating the 20-basis-point ROE reduction and vacating the $215.5m SCR disallowance, and remanding. Implemented via ACC Decision No. 78979 and the Court Resolution Surcharge.
• Burns v. Arizona Public Service Company, No. CV-21-0080-PR (Ariz. Sup. Ct., 27 September 2022) (Bolick, J.) — a Commission majority may not prevent an individual commissioner from exercising investigatory powers under Ariz. Const. art. 15 the relevant section. https://www.azcourts.gov/Portals/0/OpinionFiles/Supreme/2022/CV210080PR.pdf
• Arizona Supreme Court, September 2022 (Bolick, J., 4-1) — a single Corporation Commissioner may unilaterally subpoena a public service corporation’s political, lobbying and charitable spending records.
• Arizona Court of Appeals (2004) — voided components of the ACC’s 1999 retail-competition rules as unconstitutional.

4. Press and trade sources

• KJZZ, “New laws protect Arizona utilities from some wildfire liability, allow securitization,” 15 May 2025 — https://www.kjzz.org/politics/2025-05-15/new-laws-protect-arizona-utilities-from-some-wildfire-liability-allow-securitization
• KJZZ, “Arizona appeals court revives suit against Corporation Commission’s new yearly rate adjustments,” 25 November 2025 — https://www.kjzz.org/politics/2025-11-25/arizona-appeals-court-revives-suit-against-corporation-commissions-new-yearly-rate-adjustments
• KJZZ, “Thompson, Heap leading Republican primary for 2 open seats on Arizona Corporation Commission,” 21 July 2026 — https://www.kjzz.org/elections/2026-07-21/thompson-heap-leading-republican-primary-for-2-open-seats-on-arizona-corporation-commission
• KJZZ, “2 Arizona Corporation Commissioners ran for reelection. Only 1 made it out of GOP primary,” 23 July 2026 — https://www.kjzz.org/politics/2026-07-23/2-arizona-corporation-commissioners-ran-for-reelection-only-1-made-it-out-of-gop-primary
• KJZZ, “Extended Corporation Commission hearings on APS rate increase wrap,” 9 July 2026 — https://www.kjzz.org/business/2026-07-09/extended-corporation-commission-hearings-on-aps-rate-increase-wrap
• Arizona Capitol Times, “APS rate case kicks off with hours of protest over 14% rate increase,” 19 May 2026 — https://azcapitoltimes.com/news/2026/05/19/aps-rate-case-kicks-off-with-hours-of-protest-over-14-rate-increase/
• Arizona Capitol Times, “How would APS and TEP’s formula rate adjustments work?”, 21 May 2026 — https://azcapitoltimes.com/news/2026/05/21/how-would-aps-and-teps-formula-rate-adjustments-work/
• Arizona Capitol Times, “Attorney General Mayes sparks turf war with Arizona Corporation Commission,” 9 April 2026 — https://azcapitoltimes.com/news/2026/04/09/attorney-general-mayes-sparks-turf-war-with-arizona-corporation-commission/
• Arizona Capitol Times, “Freedom Caucus announces primary challengers for Arizona Corporation Commission,” 10 June 2025 — https://azcapitoltimes.com/news/2025/06/10/freedom-caucus-announces-primary-challengers-for-arizona-corporation-commission/
• Arizona Mirror, “APS docs reveal it funded 2014 dark money effort supporting commissioner’s son,” 1 April 2019 — https://azmirror.com/2019/04/01/aps-docs-reveal-it-funded-2014-dark-money-effort-supporting-commissioners-son/
• Arizona Mirror, “Taking stock: APS election spending, Don Brandt, millions,” 3 September 2019 — https://azmirror.com/2019/09/03/taking-stock-aps-election-spending-don-brandt-millions/
• KJZZ, “Burns steps up dark money quest, subpoenas APS and Pinnacle West,” 25 August 2016 — https://www.kjzz.org/2016-08-25/content-355927-burns-steps-dark-money-quest-subpoenas-aps-and-pinnacle-west
• California Energy Markets / NewsData, “APS focused on regulatory relationships as effects of 2021 rate case persist” — https://www.newsdata.com/california_energy_markets/financial_news/aps-focused-on-regulatory-relationships-as-effects-of-2021-rate-case-persist/article_73a7151c-ba05-11ed-9441-7fa7c84f280c.html
• Arizona State Law Journal, “The Laws of Fire in the Burning West,” 23 March 2026 — https://arizonastatelawjournal.org/2026/03/23/the-laws-of-fire-in-the-burning-west/
• The Gila Herald, “Arizona adopts new fire mitigation and utility liability law…” — https://gilaherald.com/arizona-adopts-new-fire-mitigation-and-utility-liability-law-to-help-prevent-wildfires-and-protect-ratepayers/
• Chamber Business News, “Two bills on energy infrastructure reach governor’s desk,” 8 May 2025 — https://chamberbusinessnews.com/2025/05/08/two-bills-on-energy-infrastructure-reach-governors-desk/

5. Industry benchmarking and market data

A note on two claims deliberately NOT relied on. (i) Several secondary sources report that Arizona Attorney General Kris Mayes recommended a 6% ROE in the 2025 rate case. The Attorney General’s intervention in opposition to the increase is well established, but the specific 6% figure could not be corroborated to testimony or a primary filing, and at least one source misidentified Mayes as a sitting commissioner rather than as Attorney General (she is a former commissioner). The figure is therefore excluded from the analysis. (ii) A reported RRA upgrade of Arizona within the Below Average band after the February 2024 order is plausible but paywalled and unverified; only the well-documented 2021 downgrade to Below Average is relied on.

• S&P Global Market Intelligence / Regulatory Research Associates — median authorised electric ROE 9.75% (Q1-2025) and 9.70% (FY2024), 2025 average award 9.66%. RRA lowered Arizona to Below Average, its lowest category, following the November 2021 rate case (Arizona had been Average/3 in 2016), and named Arizona among states to watch for near-term shifts in regulatory risk on commissioner turnover. Arizona’s current 2025–26 sub-tier could not be confirmed to a primary source — RRA’s state evaluations are behind an S&P Global subscription — and no upgrade is relied on in this memo. https://www.spglobal.com/market-intelligence/en/news-insights/research/underearning-spread-widens-for-gas-electric-utilities-in-roe-analysis https://www.spglobal.com/market-intelligence/en/news-insights/research/8-states-bear-watching-for-near-term-shifts-in-utility-regulatory-risk (23 January 2025) https://www.spglobal.com/market-intelligence/en/news-insights/research/fluctuations-in-regulatory-risk-have-implication-for-utility-financial-performance
• AZI price history (adjusted OHLCV, dividends, splits, EMAs, beta, alpha), full history to 24 July 2026 — https://azitrading.com/controls/download-data.php?t=PNW. Local copy output/PNW/2026-07-25/_scratch/PNW_prices.csv. Used for all price-move, drawdown and event-map arithmetic, on the adjusted close, close-to-close.
• AZI valuation index (scripts/azi.sh fundamentals, .valuation_index) as of 24 July 2026 — own-history percentile ranks for PNW and the 13-name regulated peer set (AEE, ATO, CMS, DTE, EVRG, IDA, LNT, NI, OGE, POR, PPL, WEC, XEL). Own-history context only, never cross-sectional.
• FactorsToday factor model (no-auth) — /api/stock-loadings/PNW, /api/leaderboard/PNW, /api/stock-info/PNW, /api/stock-specific-vol/PNW, /api/related-stocks/PNW, /api/factor-returns/historic. https://www.factorstoday.com/api (methodology at /about). Third-party statistical estimates; loadings and realised returns are reportable facts, persistence is interpretation. All leaderboard returns are annualised, including short windows.
• ROIC.ai MCP — get_income_statement, get_balance_sheet, get_cash_flow, get_profitability_ratios, get_enterprise_value, get_valuation_multiples, get_company_profile, get_company_news, list_earnings_calls, get_earnings_call_transcript. Third-party aggregated data, NOT primary. Documented discrepancies identified and corrected against the filing in this engagement: (i) return_com_eqy is unusable cross-sectionally for regulated utilities (returns 29.4% for AEE, 19.4% for XEL, 18.8% for WEC — impossible against ~9.5–10.5% authorisations), so peer ROE was derived independently as (P/B)÷(P/E) from a single consistent source; (ii) cf_free_cash_flow equals cash from operations and does not deduct capex; (iii) the D&A used in the EBITDA build ($969.6m) is the cash-flow line including nuclear fuel, not the income-statement figure ($915.3m); (iv) “total debt” of $14,299.6m is borrowings of $10,562.7m plus operating lease liabilities of $3,737.0m; (v) FY2024 basic EPS is $5.35 per the filing, not $5.15; (vi) tot_debt_to_tot_cap returns implausible values for some peers (85.4% for CMS).
• US Energy Information Administration state electricity price data (Arizona residential ~15.5¢/kWh against a ~18.8¢ national average; Arizona average monthly consumption ~1,075 kWh producing an above-national average bill).
• Itron Annual Energy Survey (2025) — national average residential customer growth benchmark.
• US EIA Form EIA-861, “Sales to Ultimate Customers” (calendar 2024 and prior years) — https://www.eia.gov/electricity/data/eia861/ — the consistent-basis source for the APS-versus-SRP rate comparison: CY2024 residential APS 16.45¢/kWh vs SRP 13.46¢/kWh (SRP 18.2% cheaper); all classes APS 14.53¢ vs SRP 11.59¢ (20.2% cheaper); average monthly residential bills APS $170.02 on 1,034 kWh vs SRP $162.97 on 1,211 kWh.
• Salt River Project Agricultural Improvement and Power District — 2025 Annual Report (FY ended 30 April 2025), https://www.srpnet.com/assets/srpnet/pdf/about/2025-annual-report.pdf; Preliminary Official Statement, 2025 Series C Electric System Revenue Bonds, 8 October 2025 (audited FY2025 combined financials at Appendix A), https://www.srpnet.com/assets/srpnet/pdf/doing-business/SRP-Preliminary-Official-Statement-2025-Series-C.pdf. Key figures: 1,183,163 customer accounts (1,069,185 residential); FY2025 operating revenues $4.56bn; accumulated net revenues $7.01bn; total debt $6.31bn; ratings Moody’s Aa1 / S&P AA+; 20.2% of Palo Verde (803 MW) and 10% of Four Corners (150 MW); board elected on an acreage-weighted basis.
• Agreement of August 31, 1955 (the “Territorial Agreement”) between SRP and Arizona Public Service Company — filed as Exhibit 10.1 to APS’s Form 10-Q for the quarter ended 31 March 1998, https://www.sec.gov/Archives/edgar/data/7286/0000950147-98-000400.txt — and still listed as Exhibit 10.29 to the PNW/APS FY2025 10-K, together with the Power Coordination Agreement (Exhibit 10.30), the 1998 Memorandum of Agreement (10.31) and its Addendum (10.31(a)). Provides exclusive retail distribution territories as against each other, defined by metes and bounds; 35% liquidated damages for serving into the other’s territory; survives municipal annexation.
• Ball v. James, 451 U.S. 355 (1981) — upholding SRP’s acreage-weighted board franchise as a limited-purpose special district outside one-person-one-vote. Arizona Constitution art. XIII the relevant section and art. XV the relevant section; A.R.S. the relevant section-2302; ACC Electric FAQs confirming SRP is outside ACC rate jurisdiction (https://azcc.gov/utilities/electric/electric-faqs).
• Salt River Project — corporate disclosures and rate filings (February 2025 board-approved net 2.4% increase effective November 2025); the E-67 large-load price plan (mandatory above 20 MW, 80%-of-forecast minimum billing demand); the Large Customer Integration Process; SRP High-Tech Interconnect Project.

6. Peer set used for cross-read

Twenty-four US regulated electric, gas and water utilities plus independent power producers were used as the comparison set for authorised-ROE benchmarking, rate-base and capital-plan comparison, EPS-algorithm benchmarking, data-centre contracting discipline and valuation context:

Ameren (AEE) · American Electric Power (AEP) · Atmos Energy (ATO) · CMS Energy (CMS) · CenterPoint (CNP) · Dominion (D) · DTE Energy (DTE) · Duke Energy (DUK) · Consolidated Edison (ED) · Edison International (EIX) · Entergy (ETR) · Evergy (EVRG) · Exelon (EXC) · FirstEnergy (FE) · Alliant (LNT) · NextEra (NEE) · NiSource (NI) · PG&E (PCG) · PSEG (PEG) · PPL · Southern Company (SO) · Sempra (SRE) · WEC Energy (WEC) · Xcel (XEL)

PG&E and Edison International supplied the California inverse-condemnation contrast used to assess Arizona’s H.B. 2201. Southern Company, operating before Georgia’s elected Public Service Commission, supplied the counter-example establishing that an elected commission does not by itself determine outcomes. Xcel supplied the affordability-headroom framework. All peer figures are as of each company’s most recent public disclosure in June–July 2026.

7. Analytical frameworks

• Bruce Greenwald and Judd Kahn, Competition Demystified — barriers to entry as the dominant variable; the taxonomy of genuine advantages (supply/cost, demand/captivity, economies of scale plus captivity, and government protection); the market-share-stability and ROIC tests (15–25% indicates advantages present, 6–8% indicates advantages absent).
• Edward Chancellor / Marathon Asset Management, Capital Returns — supply-side capital-cycle analysis; the asset-growth anomaly; the treatment of regulated industries as a deliberate exception in which the capital cycle’s disciplining mechanism is relocated from the product market to the political market.

8. Method and data notes

Every claim in this article traces to a primary filing, a dated public source, or an explicitly labelled derivation. The trailing five-year SEC corpus for CIK 0000764622 (431 filings) was enumerated and reviewed in full. All 202 Forms 4 and 10 Forms 144 were parsed from raw XML rather than read in summary, which is how the “exactly one open-market purchase in five years” and “no Form 4 references a 10b5-1 plan” findings were established.

Price, drawdown and event-map arithmetic uses the split- and dividend-adjusted closing series and quotes moves close-to-close; the derived all-time high was cross-checked against an independent relative-strength-to-peak measure and reconciles exactly. The APS-versus-SRP rate comparison uses EIA Form 861 for calendar 2024 on both sides, rather than mixing each utility’s own fiscal-year disclosure, because the two report on different bases.

Where a third-party data aggregator and an SEC filing disagree, the filing governs, and the six specific discrepancies found during this work are documented in the market-data section above. Two widely-circulated claims were investigated and deliberately excluded for want of primary corroboration; both are identified in that section. A third — that Arizona’s renewable-energy standard was repealed in March 2026 — is reported as settled by most trade press but is not in effect, per the company’s own quarterly filing; the report states the actual status.