CF Industries Holdings, Inc. (NYSE: CF) — The Low-Cost King, Priced for the Crisis to Last

Independent equity research note. Report date: 2026-06-19. Price referenced: $102.93 (2026-06-18 close).

This note discusses valuation only as embedded expectations and scenarios. The body (sections 1–15) contains no buy/sell recommendation and no price target. The single, deliberate exception is the Author’s Take block immediately below, which is fenced off as a subjective opinion.

⚡ Author’s Take

This block is the author’s own subjective opinion and general information only — not investment advice. The analysis that follows (sections 1–15) takes no position and carries no price target.

Verdict: HOLD / quality-cyclical-at-a-full-normalized-price / accumulate-on-weakness toward the low-$80s / not-a-short. Medium conviction.

CF is the best operator on a structurally mediocre street — the lowest-cost, largest-scale nitrogen producer in the most stable jurisdiction, run by a management team that has done the two things a cyclical’s stewards should do: build a fortress balance sheet (net debt/EBITDA ~0.4x) and retire ~33% of the share count, buying hardest at the 2024 trough and pausing the buyback during the 2026 spike. None of that is in dispute. What I will not pay full freight for is the story now wrapped around it. At $102.93 the stock embeds a partial “CF premium” — management’s claim that the Iran/Strait-of-Hormuz supply shock has permanently bifurcated the cost curve and structurally raised the mid-cycle nitrogen price. That is the narrative every commodity producer tells at the top, and the tape is already disagreeing with it: the stock is −25% from its March-2026 all-time high of $137.60 as the geopolitical premium unwinds and Henry Hub falls back toward ~$2.60. The “cheap” 9x P/E (18.6th percentile of its own history) is the classic peak-earnings mirage — the cycle-immune P/B (69.9th) and P/S (68.8th) say CF is in the upper third of its own decade range. On a normalized mid-cycle EBITDA of ~$2.8–3.2B (vs ~$3.5B TTM, which carries a one-time ~$170M litigation gain), ~6–7x EV/EBITDA is fair-to-full, not a bargain.

So this is a great business at a fair-to-full normalized price, framed correctly as a quality cyclical at a geopolitical-premium top, not a value setup and not a falling knife (the multi-year uptrend is intact; FactorsToday shows it trading as a defensive, high-alpha energy instrument — OilPrice β~1.2–1.32, market β only 0.165). I would own it lower. The math: ~6x on ~$2.8–3.2B mid-cycle EBITDA implies an EV around $17–19B and equity roughly $83–95/share — my accumulation zone is the low-to-mid-$80s, where you are paid the cost-curve advantage and the buyback compounding without underwriting the unproven premium. Conviction: medium. Flips bullish if nitrogen holds materially above the pre-2022 norm for multiple quarters despite a calm geopolitical backdrop and resumed Chinese exports (proving the higher incentive price is structural, not scarcity-driven). Flips bearish if Chinese urea exports resume and NOLA reverts toward $300–350/st while management leans further into cycle-top growth capex. Tag: the best house on the cost curve — renting a premium the market is already handing back.

📈 Stock Price Action — Five-Year Event Map

Factual price history — not a recommendation, not a price target.

The arc (FACT, AZI price CSV, 2026-06-18; split/dividend context per the unadjusted series). Over the trailing five years CF has run a full commodity round-trip and then printed a fresh all-time high. The stock bottomed near $38.53 (Jan-2021) in the post-COVID nitrogen trough, ripped to ~$119–120 in 2022 on the European energy crisis (Russian/Ukrainian gas-feedstock destruction), spent 2023–2025 range-bound in the high-$70s/$80s as nitrogen normalized, then spiked to an all-time high of $137.60 on 2026-03-30 on the Iran / Strait-of-Hormuz nitrogen supply shock. It now trades at $102.93 (2026-06-18), roughly -25% off the March ATH but still ~2.7x the 2021 low. The 52-week range is $76.09–$137.60 — i.e., the stock has more than doubled trough-to-peak inside twelve months, the signature of an energy-levered commodity instrument (FactorsToday: dominant factor = OilPrice β~1.2–1.32), not a stable compounder.

Cycle narrative. (1–2) The 2021–2022 surge was a textbook capital-cycle/commodity squeeze: war-driven destruction of European gas-fed ammonia capacity handed North American low-cost producers a windfall, lifting CF’s EBITDA to $6.4B and EPS to $16.39 in 2022 — the peak the rest of the cycle is measured against (FACT, 10-K; INTERP on attribution). (3–4) As energy normalized, earnings mean-reverted (EBITDA $3.1B → $2.6B) and the stock spent two years range-bound; management used the lull to retire ~33% of shares since 2017 and bolt on Waggaman (FACT). (5) The 2026 melt-up to a fresh ATH was a geopolitical supply shock, not a demand or structural-margin event: management cited 31 Middle-East ammonia plants impacted, 49 South-Asian plants curtailed on LNG feedstock, and 20–21 drone-hit Russian plants (FACT, Q1’26 call 2026-05-07) — a textbook one-quarter scarcity spike. (6–7) The ~-25% retrace since March tracks the unwind of that risk premium as Henry Hub fell back to ~$2.60 and US-Iran tensions eased — consistent with the FactorsToday read that this is an OilPrice-beta instrument experiencing a risk-premium normalization, not a thesis-breaking falling knife (INTERP).

1. Executive Summary

CF Industries is the world’s largest publicly traded nitrogen producer — a pure-play converter of cheap North American natural gas into globally priced ammonia, urea, UAN and ammonium nitrate, distributed through an owned Corn Belt logistics network. It is a clean, understandable, single-commodity business with three genuine strengths and one inescapable structural limitation.

The strengths. First, a real Greenwald supply/cost advantage: privileged access to structurally cheap, abundant U.S. shale gas (Henry Hub ~$3/MMBtu vs European TTF ~$16, ~5x) plus the scale of Donaldsonville — the largest ammonia complex in the world — places CF firmly in the first quartile of the global cost curve, earning a fat spread over the European marginal producer that sets the price. Second, a fortress balance sheet (net debt/EBITDA ~0.4x, ~22x interest coverage, no maturity wall before 2034, recently upgraded from secured to all-unsecured notes) — exactly the posture a 2.4x-EBITDA-cyclical needs. Third, best-in-class, counter-cyclical capital allocation: a ~33% share-count reduction since 2017 via buybacks concentrated at the cyclical trough, a deliberately modest and well-covered dividend (~15–22% payout), and a well-timed, below-replacement-cost brownfield acquisition (Waggaman, Dec-2023, ~$1.675B).

The limitation. CF has no demand-side moat — no customer captivity, no switching costs, no pricing power. Its own 10-K concedes the industry has “limited barriers to entry” and competes “primarily on delivered price.” It is a price-taker on both its output (global nitrogen benchmarks) and its dominant input (natural gas); profit is the arbitrage between the two. The numbers prove it: EBITDA swung from $1.49B (2020) to $6.42B (2022, the European-energy-crisis blowout) back to $2.64B (2024) and $3.27B (2025); ROIC oscillated from 5.7% (below WACC, at the trough) to 41.3% (at the peak). Revenue moves entirely on price; sales volume is dead flat at ~19M tons. This is a cost-advantaged commodity producer, not a compounding franchise — own it for cost-curve position and capital returns through the cycle, not for durable excess returns.

The situation today. A late-Q1’26 Iran conflict and Strait-of-Hormuz closure delivered the third major nitrogen supply shock in six years, spiking global prices and lifting CF to an all-time high of $137.60 (Mar-2026). Management has reframed the windfall as a structural re-rating — the “CF premium,” in which geopolitics permanently bifurcates the first quartile into “low-cost-and-low-risk North America” vs “fragile, exposed” Middle East/Russia. The stock has since retraced ~25% to $102.93 as the premium unwinds (US-Iran de-escalation; Henry Hub back to ~$2.60). The central question for any buyer is whether ~$3.5B TTM EBITDA is the new mid-cycle (the bull/“CF premium” case) or a near-peak that mean-reverts toward ~$2.5–3.0B (the bear/Marathon capital-cycle case). On P/E and FCF yield (~10–11%) CF screens cheap; on EV/EBITDA (~5.8x spot), P/B, P/S and a normalized-EBITDA lens it is fair-to-full. The market is underwriting a partial premium — a mid-cycle above the pre-2022 norm but below the 2026 spike. That embedded bet, against a no-moat commodity whose capital cycle is currently adding low-cost Gulf Coast supply (including CF’s own Blue Point JV, +1.5M tons by 2029), is the crux this memo dissects.

2. Business Overview

What CF makes — the ammonia value chain

FACT. At its core CF Industries is a producer of anhydrous ammonia (NH₃, 82% nitrogen / 18% hydrogen). It uses the century-old Haber–Bosch process to fix atmospheric nitrogen with hydrogen stripped from natural gas (steam methane reforming, SMR). Natural gas is therefore both the chemical feedstock and the energy source — the single fact that governs the entire economics of the business. From ammonia, CF either sells the molecule directly or upgrades it into higher-value nitrogen derivatives (Source: CF 10-K FY2025, Item 1 “Our Strategy,” filed 2026-02-25):

• Granular urea (46% N) — the most-traded solid nitrogen fertilizer globally; also the feedstock for DEF and urea liquor.
• UAN (urea ammonium nitrate solution, ~28–32% N) — liquid fertilizer, the most logistically “captive” product (heavy, dilute, expensive to ship — a regional product).
• Ammonium nitrate (AN) — fertilizer and a commercial-explosives input (mining/construction).
• Industrial / specialty derivatives — diesel exhaust fluid (DEF, urea-based), urea liquor, nitric acid, aqua ammonia; sold to industrial rather than agricultural customers.

FACT. CF reports five product segments — Ammonia, Granular Urea, UAN, AN, and Other. FY2025 segment net sales and gross margin (10-K FY2025, segment note):

INTERPRETATION. The mix is instructive: the upgraded products (urea, UAN) carry materially fatter gross margins (43–47%) than raw ammonia (31%) and the explosives-grade AN (19%). The value chain — turning a $-per-MMBtu gas molecule into a $372/ton-average nitrogen product (FY2025 realized; +19% YoY from $313 in 2024, 10-K MD&A) — is where the economics sit. Volume is remarkably flat (19.1M / 18.9M / 19.1M product tons in 2023/2024/2025) — this is a price story, not a volume story (10-K). Production volumes by product 2025: ammonia 10.12M tons (gross), granular urea 4.26M, UAN 6.93M, AN 1.25M.

Recurring vs. commodity/spot revenue

FACT / INTERPRETATION. There is essentially no recurring revenue in the SaaS sense. CF is a price-taker selling globally-traded commodities; revenue = volume × spot-linked price. CF’s own 10-K states its markets are “intensely competitive, based primarily on delivered price.” The CHS supply agreement (below) and some forward-priced industrial DEF/AN contracts provide modest visibility, but ~all revenue re-prices each cycle to the global nitrogen benchmark. Realized prices swung from $207/ton (2017) to a 2022 energy-crisis peak, back to $313 (2024), to $372 (2025) — a fundamentally cyclical, externally-determined revenue line.

Geographic & asset footprint — the world’s largest ammonia network

FACT (10-K FY2025, Item 1 & Item 2 Properties):

• Six U.S. complexes: Donaldsonville, LA (the largest ammonia complex in the world, ~40% of CF’s ammonia capacity); Port Neal (Sergeant Bluff, IA); Yazoo City, MS; Verdigris (Claremore, OK); Woodward, OK; and Waggaman, LA (acquired Dec-2023).
• Two Canadian complexes: Medicine Hat, Alberta (largest ammonia complex in Canada) and Courtright, Ontario.
• One UK complex: Billingham (Ince was closed; UK operations restructured — $23M restructuring charge in 2025).
• JV interests: 50% of Point Lisas Nitrogen Ltd (PLNL) in Trinidad (equity method, gas-curtailment-prone); 40% of the new Blue Point JV (consolidated as a VIE).
• Distribution: an “extensive system of terminals and associated transportation equipment located primarily in the Midwestern United States” — the in-house logistics network into the Corn Belt (rail, barge, pipeline, terminals). FY2025 consumed ~350 million MMBtu of natural gas across the fleet.

The CF Nitrogen LLC / CHS strategic-equity structure

FACT. Five of the six U.S. plants are held inside CF Industries Nitrogen, LLC (CFN), of which CF owns ~89% and CHS Inc. (the large U.S. ag cooperative) owns ~11% of the membership interests. The Waggaman plant sits outside CFN (wholly owned). The structure dates to a 2016 transaction in which CHS paid CF ~$2.8B for the equity stake plus a long-term offtake right. Under the supply agreement, CHS may purchase up to ~1.1M tons of granular urea and ~580,000 tons of UAN annually from CFN at market prices, and receives semi-annual cash distributions proportional to its equity interest. CHS was CF’s largest customer in 2025 at ~13% of consolidated net sales (10-K, “Customers”).

INTERPRETATION — how the minority interest flows through. Because CF consolidates CFN, ~11% of CFN’s net earnings are stripped out as noncontrolling interest (NCI) before arriving at “net earnings attributable to common stockholders.” On the balance sheet the CHS stake shows as ~$2.94B of minority interest within total equity (company filings). This matters two ways for analysis: (1) the NCI deduction is pro-cyclical — at peak nitrogen prices CHS takes a bigger absolute slice, so reported EPS is more cyclically damped than gross EBITDA suggests; (2) for enterprise value, the ~$2.94B NCI must be added to the bridge (as it was: live EV ~$20.2–20.5B), materially larger than net debt (~$1.2B). ASSUMPTION: the CHS arrangement is effectively a permanent quasi-debt/quasi-equity claim on ~11% of the core U.S. earnings stream — not a strategic partner CF can easily buy back cheaply at the top of a cycle.

OPEN QUESTION. What is the effective cost of the CHS structure to common holders over a full cycle versus simply having retained 100% and financed Waggaman-style? The semi-annual distributions are real cash leakage at peaks.

Customers & end markets — ag vs. industrial

FACT / INTERPRETATION. Principal customers are “cooperatives, retailers, independent fertilizer distributors, traders, wholesalers and industrial users” (10-K). The large majority of volume is agricultural (fertilizer — nitrogen provides crop energy), with a smaller, higher-margin industrial tail (DEF, AN for explosives, nitric acid, urea liquor for emissions control). CF does not break out a precise ag/industrial split, but the product mix and customer list imply roughly three-quarters-plus agricultural by tonnage. The business is seasonal — peak demand at North American spring planting, a second peak post-fall-harvest — which is why in-region storage/logistics (the terminal network) is a genuine operating asset, letting CF produce year-round and place tons into the seasonal window.

Verdict. CF is a pure-play, vertically-integrated nitrogen manufacturer converting cheap North American natural gas into globally-priced fertilizer and a smaller industrial-chemicals stream, distributed through an owned Corn Belt logistics network. It is a clean, understandable, single-commodity business — no conglomerate noise — but it is fundamentally a price-taking commodity producer whose revenue is volume-flat and price-cyclical, with ~11% of its core U.S. earnings owned by CHS and stripped out as minority interest. The clarity of the model is a positive for analysis; the absence of recurring revenue or pricing power is the defining structural fact the rest of this memo must price.

3. Industry Dynamics

Industry structure — fragmented, commodity, no demand-side concentration

FACT. Global nitrogen (ammonia/urea/UAN/AN) is a fragmented, intensely competitive commodity industry. China is the largest single producing nation; the top-five merchant suppliers (Yara, CF, OCI [now winding down], SABIC Agri-Nutrients, Nutrien) together hold only a limited share of global capacity (peer-research sweep; Grand View Research). Urea — the globally-traded benchmark product — is the least concentrated; UAN is somewhat more regional/consolidated because it is heavy and dilute (uneconomic to ship far). CF’s own 10-K is unusually blunt: ammonia, urea and UAN are “widely-traded fertilizer products and there are limited barriers to entry” (Item 1, Competition). A producer rarely volunteers that there are limited barriers to entry; take it at face value.

The cost curve — the entire game is natural-gas feedstock

FACT. Natural gas is ~70–90% of the cash cost of nitrogen production globally (industry-standard range; Argus, AGA; confirmed in peer sweep). The global nitrogen price is set by the marginal (highest-cost) producer needed to clear demand — and CF’s 10-K explicitly frames profitability this way: prices are determined by “the industry’s marginal producers.” Because the only large variable cost is gas, the global cost curve is essentially a map of regional gas prices. This is why CF earns its position:

(Source: Global LNG Hub, 2026-06-15; EIA STEO.) FACT. Henry Hub averaged ~$2.21/MMBtu in 2024 (an inflation-adjusted record low), ~$3.52 in 2025, with EIA forecasting ~$4.0–4.4 for 2026. CF’s realized gas cost (incl. derivatives) was $3.31/MMBtu in 2025 (+38% YoY from $2.40), and management quoted ~$2.60 spot at the Q1’26 call (company filings). INTERPRETATION. US producers sit in the first quartile of the global cost curve; European gas-based producers (Yara’s European fleet) sit at the high/marginal end — they set the price, CF earns the spread. The ~5x current US-vs-Europe gas ratio is wider than the more normal ~3–4x (partly a 2026 conflict-shock distortion) but the direction and persistence of the gap is the structural fact. US shale keeps Henry Hub structurally low; CF effectively monetizes stranded North American gas as exportable nitrogen.

The cyclicality mechanism — the gas-nitrogen spread

FACT / INTERPRETATION. CF’s margin is the spread between the globally-set nitrogen price (driven by the marginal European/Asian producer’s high gas cost) and CF’s own low Henry Hub gas cost. This spread is violently cyclical because the two legs move independently:

• When global gas spikes (2021–22 European energy crisis), the marginal cost — and therefore the nitrogen price — soars while CF’s Henry Hub cost rises far less. Result: gross margin 19% (2020) → 52% (2022), EBITDA $1.49B → $6.42B, ROIC 5.7% → 41.3% (company filings). A textbook windfall.
• When global gas normalizes and/or new low-cost supply lands, the spread compresses: 2023–24 saw EBITDA fall back to ~$2.6–3.1B, ROIC to ~13–16%.

This is the single most important thing to internalize: CF’s earnings are a leveraged play on the global-vs-US gas spread, which is why FactorsToday finds CF’s dominant statistical factor is OilPrice (beta ~1.2–1.32) and its comp set is energy names (XOM, OXY, DVN), not ag-inputs (company filings). CF trades as a commodity-energy instrument.

Demand drivers — steady, inelastic, low-growth

FACT. Demand is anchored to food: nitrogen is consumed applying fertilizer to crops, with corn the largest N-consuming crop. US corn planted acreage: ~90.6M (2024) → 95.2M (2025, near-record) → ~95.3M intended (2026) (USDA NASS). Global fertilizer-use grows only ~1–2% per year (IFA medium-term outlook to 2029). INTERPRETATION. Demand is the boring, stable side of the equation — inelastic (people eat), low-growth, and not the investment swing factor. The swing factor is always supply/cost, not demand — exactly the Marathon “focus on supply, not demand” tenet. A modest 2026 acreage shift toward soybeans (lower N requirement) is a marginal headwind, not a thesis-changer.

Regulation — a genuine emerging tailwind for low-carbon, North American producers

FACT. Three regulatory vectors, all currently favoring CF’s positioning:

• US Section 45Q carbon-capture credit: $85/metric ton CO₂ for point-source capture with geologic sequestration (locked through 2026, inflation-adjusted after). CF’s Donaldsonville CCS (live July 2025, up to 2M t/yr CO₂) directly qualifies; a sell-side estimate pegged potential CCS-credit EBITDA at ~$50M/quarter (Benzinga/analyst). The 2025 OBBBA retained credit values but added foreign-entity restrictions (Payne Institute) — which, if anything, advantages a US producer vs. foreign-influenced entities.
• EU CBAM (Carbon Border Adjustment Mechanism): definitive phase began 1 Jan 2026; fertilizer/ammonia covered; certificate purchases begin ~Feb 2027. CBAM penalizes high-carbon-intensity imports (Russian/Chinese coal-route ammonia) and advantages low-carbon, gas-route + CCS producers exporting into Europe (ICAP; Argus). CF made its first low-carbon ammonia premium sales into Europe/Africa in 2025.
• Environmental: standard heavy-industry air/GHG permitting; the Verdigris nitric-acid abatement project (Q4’25) cut CO₂e >600,000 t/yr.

Supply additions — the capital-cycle read (Marathon lens)

FACT. New nitrogen capacity is coming, not starved — moderate but real:

• Global N-effective capacity +4% (166.3 Mt 2024 → 172.7 Mt 2026); global ammonia capacity ~240 Mt (2023) → ~290 Mt by 2030, with ~113 announced ammonia plants by 2028 (peer sweep; USGS; Offshore Technology).
• North American Gulf Coast is a hotspot: Ascension Clean Energy (~7.2 Mt/yr, ~2027), St. Charles Clean Fuels (~3 Mt/yr, ~2027), CF Blue Point (~1.5 Mt, 2029) — much of it clean/blue ammonia tied to energy-transition offtake.
• FSU (Nakhodka, Kemerovo), Qatar (QAFCO 6.4 Mt blue complex), Saudi (SABIC Jubail +2.6 Mt urea) all adding.

INTERPRETATION (capital cycle). This is net-positive supply growth into the late-2020s, much of it from the cheapest gas regions (US Gulf, Middle East, FSU). By Marathon logic, high returns (the 2022 windfall, the 2026 conflict spike) are attracting capital — the canonical late-cycle warning. Much new tonnage is blue/clean ammonia whose merchant-urea impact depends on whether energy-transition offtake actually materializes (it is slipping — see the relevant section); if that demand disappoints, the molecules spill into the merchant pool and compress margins. This is the central capital-cycle risk to the bull case. Offsetting it: project lead times are long (4–6 years), permitting is slow, and a chunk of capacity is earmarked (not merchant). The cycle is not as supply-frothy as 2015–17 (when prices fell from $314 to $207/ton on a capacity wave the 10-K still references), but it is decidedly not a supply-starved setup.

The “geopolitical risk premium” / bifurcation thesis — assess critically

FACT. Management’s Q1’26 narrative: an active Iran conflict + closure of the Strait of Hormuz disrupted a large share of low-cost supply — ~31 ME ammonia plants impacted, India/Pakistan/Bangladesh LNG-feedstock curtailments, Russian plants drone-hit, Egypt added a $90/t export duty, China restricting exports (company filings). Independent data corroborate the event: ~30% of global urea trade comes from Iran + Hormuz-constrained countries; Russia+Belarus ~14–16% of urea, ~23% of ammonia; urea spiked +26% week-on-week (decade record), Middle East FOB to ~$700+/t (peer sweep; Rystad/CRU/Bloomberg). Management’s structural claim (“CF premium”): geopolitical risk has bifurcated the first quartile of the cost curve into low-cost-AND-low-risk (North America) vs. low-cost-but-fragile (ME/Russia, ~half of first-quartile capacity), which it argues permanently raises the mid-cycle incentive price.

INTERPRETATION — skeptical. The event is real and CF is genuinely advantaged by being low-cost in a stable jurisdiction. But the structural permanence claim is a classic peak-narrative-building exercise and should be treated as a hypothesis, not a fact (management commentary is a hypothesis, not evidence). Three reasons for skepticism: (1) the same Middle East/Russian capacity has weathered prior shocks and returned — Hormuz has never stayed closed; the stock already faded from the $137.60 ATH to ~$103 as US–Iran de-escalation progressed (company filings), i.e., the market is un-pricing the premium even as management institutionalizes it; (2) elevated mid-cycle prices are precisely what attracts the new low-cost Gulf/US capacity documented above — the capital cycle answers high prices with supply; (3) a “risk premium” is a rent, and rents in a no-barriers-to-entry commodity get competed away. The low-COST advantage is durable (gas geology). The low-RISK premium is a transient, narrative-dependent rent. Distinguishing the two is the crux of the whole report.

Verdict (see Industry Dynamics). Structurally a mixed-to-poor industry, with one redeeming feature. It is a fragmented, no-demand-moat, price-taking commodity with explicitly “limited barriers to entry,” steady low-growth demand, violent cyclicality, and a capital cycle currently adding supply into elevated prices — all negatives. The single redeeming structural feature is real, persistent supply-side cost dispersion driven by regional gas geology, which hands first-quartile (US/Canada) producers a durable cost advantage and a fat through-cycle spread over the European marginal producer. The regulatory vector (45Q, CBAM) is a genuine emerging tailwind for low-carbon North American producers. But the industry confers no pricing power on anyone — it is a good place to be the low-cost operator and a terrible place to be anyone else.

4. Competitive Position

Name the moat — a cost (supply) advantage, not a demand moat

INTERPRETATION (Greenwald taxonomy). CF has exactly one genuine competitive advantage, and it is a supply/cost advantage in Greenwald’s framework — privileged access to a cheap input (stranded North American natural gas) plus economies of scale in production and an owned distribution network. It has none of the demand-side advantages:

• No customer captivity / habit: buyers (co-ops, distributors, traders) purchase on delivered price; ammonia/urea/UAN are fungible commodities. Switching cost to a rival molecule of identical spec is ~zero.
• No switching costs / search costs: a ton of granular urea is a ton of granular urea.
• No network effects, no brand pricing power. CF’s own 10-K: competition is “based primarily on delivered price.”

In Greenwald’s taxonomy the cost advantage is the weakest and most transient of the three genuine types (“in the long run everything is a toaster”). The Haber–Bosch process is 100+ years old and available to all; the technology confers no edge. CF’s edge is not proprietary process — it is location + scale + logistics: cheap feedstock geology it doesn’t own but can buy at Henry Hub, the world’s largest single ammonia complex (Donaldsonville, scale economies in fixed-cost absorption), and an owned Corn Belt terminal/transport network that lowers delivered cost into the highest-value demand region.

Is the gas-cost advantage durable — or competed away?

INTERPRETATION. This is the key durability question, and the answer is the cost advantage is durable but the rent it earns is cyclical and partly competable:

• Durable element: US shale gas is structurally abundant and cheap; the US-vs-Europe/Asia gas gap (~3–5x) has persisted for over a decade and is grounded in physical geology + LNG-export bottlenecks, not a temporary dislocation. As long as Henry Hub sits well below TTF/JKM, CF’s relative position in the first quartile is secure. This passes Greenwald’s “think local / grounded in local circumstances” test.
• Competable element: the advantage is shared with every other North American producer (Nutrien’s nitrogen, Koch, CVR, LSB) and with Middle East/Russian producers who have gas as cheap or cheaper. CF is a low-cost producer, not the uniquely low-cost producer. And the high returns the advantage throws off in good years attract new low-cost capacity (Gulf Coast build-out, the relevant section) — the Marathon mean-reversion mechanism. The advantage doesn’t disappear, but the excess return on it compresses as capital responds.

The ROIC test (Greenwald) — cyclical, not consistently elite

FACT. Greenwald’s profitability test for a moat is sustained after-tax ROIC of 15–25%+ over a decade. CF’s ROIC (ROIC.ai): 5.7% (2020) / 18.2% (2021) / 41.3% (2022 peak) / 16.2% (2023) / 12.8% (2024) / 17.1% (2025) (company filings). INTERPRETATION. This is the signature of a cyclical cost-advantaged commodity producer, not a franchise. ROIC is spectacular at the top (41%) and sub-cost-of-capital at the trough (5.7% in 2020). The trough ROIC dipping toward/below WACC is the tell that there is no durable franchise insulating returns from the cycle. Average through-cycle ROIC is probably low-to-mid teens — adequate but not elite, and entirely explained by cost-curve position rather than any demand moat. Market-share stability is high (CF holds ~40–44% of North American ammonia/urea/UAN capacity, stable), which Greenwald would read as evidence of some barrier — but in a globally-traded commodity, regional production share does not equal pricing power; imports cap domestic prices continuously (10-K: “we experience competition from foreign-sourced products continuously”).

Direct competitive comparison

FACT / INTERPRETATION (peer sweep + 10-K named competitors):

INTERPRETATION. CF’s defensible claim is not “lowest cost on earth” (the Gulf and Russia are cheaper on gas) but “low cost in a stable, CBAM-favored, logistically-integrated jurisdiction, at the largest scale.” That is a real and unusual combination. Versus Yara (the European marginal producer), CF’s advantage is stark and durable. Versus Gulf/Russian producers, CF trades a few dollars of gas cost for political stability, jurisdiction, and (increasingly) carbon-intensity advantage under CBAM.

Low-COST (real) vs. low-RISK (“CF premium” — newer, weaker)

INTERPRETATION. Management is trying to upgrade the story from “we are low-cost” (true, durable, structural) to “we are low-cost and low-risk, which permanently re-rates our mid-cycle earnings” (a 2025–26 narrative). The first is a competitive-advantage fact grounded in geology and logistics. The second is a cyclical rent dressed up as a structural moat — it exists only while geopolitical disruption persists, it is being competed against by the very Gulf/US capacity additions it incentivizes, and the market has already begun fading it (stock −25% off the conflict-driven ATH). An analyst should bank the low-cost advantage and heavily discount the low-risk premium.

Verdict. A durable cost advantage in a no-demand-moat commodity — a transient-rent generator, not a franchise. CF possesses a genuine Greenwald supply/cost advantage (cheap stranded US gas + Donaldsonville-scale + owned Corn Belt logistics + a stable, CBAM-favored jurisdiction) that is durable in relative terms and gives it a structural edge over the European marginal producer that sets the price. But it confers no pricing power, no customer captivity, and no protection of trough returns — ROIC collapses to single digits at the cycle bottom, the hallmark of a cost-advantaged commodity producer rather than a franchise. The advantage is shared with other North American and Gulf producers and is competed against by new low-cost capacity whenever returns spike. The “low-cost” claim is real and bankable; the newer “low-risk CF premium” claim is a cyclical rent, not a moat. Verdict: commodity producer with a real but transient/cyclical cost edge — own it for the cost-curve position and the cycle, not for any durable franchise.

5. Growth History & Forward Opportunities

Historical growth — price, not volume; cyclical, not compounding

FACT. Revenue arc ($M): 2019 4,590 / 2020 4,124 / 2021 6,538 / 2022 11,186 (energy-crisis blowout) / 2023 6,631 / 2024 5,936 / 2025 7,084 (company filings). Against that, sales volume is dead flat: 19.1M / 18.9M / 19.1M product tons in 2023/24/25; production is similarly flat-to-modestly-up (ammonia 9.5M→10.1M tons 2023→25 as Waggaman and reliability added a bit). INTERPRETATION. Decomposing revenue (price × volume): essentially all of the variance is price/cyclical, none is organic volume growth. The 2022 doubling was the European gas crisis blowing out the gas-nitrogen spread, fully reversed by 2024. This is not a compounding business — it is a flat-volume cyclical whose top line oscillates with a spread it does not control. Where volumes have grown, it has been acquired, not organic: the 2010 Terra deal, the 2016 CHS/CFN transaction, and the Dec-2023 Waggaman acquisition (~$1.675B, a fully-built ammonia plant — capex-led, not organic).

Forward growth vector 1 — Blue Point JV (capacity-led)

FACT. Blue Point (Modeste, LA), formed April 2025 with JERA (35%) + Mitsui (25%), CF 40%. An autothermal-reforming (ATR) + CCS low-carbon ammonia plant: ~1.4M metric tons (~1.5M short tons) nameplate, capturing >95% of CO₂, online late 2029, construction starting 2026. Total facility cost ~$3.7B (ex-ASU); CF also investing ~$550M in owned scalable infrastructure (storage/vessel loading) it will operate for a fee. 2025 capital contributions: CF $195M / JERA $170M / Mitsui $121M. CF, JERA and Mitsui each off-take pro-rata to ownership (10-K, “Blue Point joint venture”). INTERPRETATION. This is capacity-led growth, ~+8% to CF’s net ammonia base by 2030 — real, but (a) four-plus years out, (b) heavily de-risked by partner offtake (JERA/Mitsui take 60%), which is also its limitation: the upside accrues largely to partners, and CF carries the construction/execution risk (a $3.7B project, with ~⅓ of cost in imported materials exposed to tariffs/timing). It is a good low-risk growth project, but it is incremental volume on a flat base, not a step-change in earnings power, and it lands into the late-2020s Gulf Coast supply wave (see Industry Dynamics).

Forward growth vector 2 — decarbonization / low-carbon ammonia optionality

FACT. Donaldsonville CCS live (July 2025, up to ~1.9M t/yr low-carbon ammonia capacity, ~2M t/yr CO₂ sequestration, ExxonMobil transport/storage, qualifies for 45Q $85/t); Yazoo City CCS ~2028 (~500k t/yr CO₂); first low-carbon premium sales into Europe/Africa in 2025; CBAM tailwind; deals/discussions with Pepsi, POET (10-K). Japan’s JERA/Mitsui committed Blue Point volumes for power generation and steel.

INTERPRETATION — assess hype vs. reality. Split into two buckets:

• Real and near-term: the 45Q credit monetization on existing Donaldsonville CCS is concrete cash (a sell-side estimate ~$50M/quarter EBITDA), and the CBAM-driven low-carbon premium into Europe is a genuine, regulation-created margin uplift on a slice of volume. Bank these as modest, real, incremental margin — not transformational, but accretive optionality with little incremental capex (the CCS units cost ~$200M / ~$100M, already spent/committed).
• Largely hype on a 2030 horizon: the clean-ammonia-as-marine-fuel and Japanese power co-firing TAM is the headline narrative and it is substantially overstated. Independent analysis (Chemical Market Analytics) finds co-firing economics “do not currently make for good economics” without subsidy; South Korea’s first clean-hydrogen auction drew bids for only 11% of volume offered; Japan’s CfD has limited participation; storage infrastructure is a fraction of targets; CMA projects Japan+Korea power-gen clean-ammonia demand reaches only ~12 Mt by 2050 vs. ~30 Mt+ government targets (industry-macro sweep). Treat clean-ammonia energy demand as a long-dated call option dependent entirely on government subsidy, not a base-case 2030 demand pillar. Management’s framing (“accelerate the world’s transition to clean energy”) is aspirational and should be heavily discounted.

Forward growth vector 3 — M&A / brownfield (capacity-led, lumpy)

FACT/INTERPRETATION. Waggaman (Dec-2023) is the template: buy or build incremental low-cost tons rather than grow organically. Industry consolidation is favorable on the supply side (OCI’s exit, Koch’s Wever purchase reduce the merchant-seller count — a mild Marathon positive). But CF’s growth path is inherently lumpy, capacity-led, and capital-intensive — there is no organic compounding flywheel because there is no pricing power or unit-economics improvement with scale beyond what already exists.

Verdict (see Growth). Low-quality growth in the value-creation sense — capacity-led and cyclical-price-led, not compounding. Historical “growth” is almost entirely cyclical price oscillation on a flat ~19M-ton volume base, with the genuine volume additions coming from acquisitions (Terra, CHS/CFN, Waggaman). Forward growth is (1) Blue Point — real but incremental (~+8% volume), four years out, de-risked-but-capped by partner offtake, and exposed to execution and the late-2020s supply wave; (2) decarbonization — split between real, modest, low-capex near-term cash (45Q credits + CBAM premiums) and largely-hyped, subsidy-dependent long-dated clean-energy TAM that should not anchor a base case; and (3) lumpy M&A/brownfield. None of it changes the fundamental character: CF grows earnings power mainly by adding low-cost tons and by riding the gas-nitrogen spread, not by compounding superior unit economics. The growth is defensible and competently-directed but structurally low-quality — own CF for cost-curve position and capital returns, not for a growth-compounding thesis.

6. Financial Quality

The core question: is this an economic engine, or a spread?

CF’s reported numbers look superb — FY2025 gross margin 38.5%, EBITDA margin ~46%, ROE 37%, ROIC ~17%, net debt/EBITDA 0.38x. But the central analytical fact about CF is that none of these are structural; they are a snapshot of where the nitrogen-vs-natural-gas spread happened to sit in 2025. CF is a price-taker on output (global urea/ammonia/UAN prices, set at the marginal high-cost producer, largely European/Asian gas-based plants) and a price-taker on its dominant input (North American Henry Hub natural gas). Its profit is the arbitrage between the two, scaled by ~19M product tons of volume. Economics do not improve with scale in the franchise-value sense; they improve with the spread. The job of this section is to size that spread, separate it from genuine durable advantage, and strip the one-time noise out of the 2025/Q1’26 prints.

Multi-year revenue / margin / EBITDA arc — the cycle is the story

FACT. The 2022 print is an energy-crisis blowout: Russia’s invasion of Ukraine spiked European/Asian gas to multiples of Henry Hub, blowing the global nitrogen cost curve out and handing CF — sitting at the low-cost North American end — a 57% EBITDA margin and $6.4B of EBITDA. By 2024 the spread had normalized and EBITDA more than halved to $2.6B. 2025’s $3.3B is a partial, mid-cycle recovery, not a new plateau. (Source: FY2025 10-K MD&A; FY2022/2023 10-Ks; ROIC.ai credit ratios, accessed 2026-06-19.)

INTERPRETATION. Peak-to-trough EBITDA swing of 2.4x ($6.4B → $2.6B) in two years is the single most important fact for valuation. Any multiple applied to a single year’s earnings is meaningless without specifying where in the cycle that year sits. 2025 is roughly mid-cycle.

Unit economics: the nitrogen margin = (output price − gas cost), quantified

The 10-K gives the exact gas-cost mechanics:

• FACT. Henry Hub averaged $3.53/MMBtu in 2025 (vs $2.25 in 2024, $2.53 in 2023). CF’s realized gas cost in COGS — including transport and realized derivatives, FIFO — was $3.31/MMBtu (2025), up 38% from $2.40 (2024). (FY2025 10-K, “Natural gas supplemental data.”)
• FACT. CF’s plants consumed ~350 million MMBtu of gas in 2025. (10-K Item 1.)
• FACT — gas-cost sensitivity (the key disclosure). Per the 10-K: “A $1.00 per MMBtu change in the price of natural gas would change the cost to produce a ton of ammonia, granular urea, UAN (32% N) and AN by approximately $32, $22, $14 and $16, respectively.” (FY2025 10-K, Item 7A.)

INTERPRETATION (sizing the company-level sensitivity). ~350M MMBtu/yr × $1.00/MMBtu ≈ ~$350M of pre-tax cost (≈EBITDA) per $1/MMBtu move in gas, holding output prices and volumes constant. That is ~11% of 2025 EBITDA from a $1 gas move — and gas swung from a $2.60 spot at the Q1’26 call to a >$7 February settle to a $6.32 average in early-2026 (10-K notes Jan 1–Feb 20, 2026 Henry Hub averaged $6.32). The earnings are violently gas-sensitive. The offsetting fact is that gas and global nitrogen prices are positively correlated (high global gas lifts the marginal cost-setter and thus output prices), which is precisely why CF mints money when European gas spikes while its own Henry Hub gas stays cheap — the 2022 dynamic. The margin is the spread, not the level. CF’s durable edge, if any, is structurally cheap, abundant North American shale gas vs. a marginal global producer on LNG-linked gas — a genuine cost advantage (Greenwald supply-side), but one that is geological/locational, not a franchise, and that fully accrues only when the global cost curve is steep.

OPEN QUESTION. Management’s Q1’26 “CF premium” narrative — that geopolitical fragility (Iran/Hormuz, Russia drone strikes, Egypt export duty) has permanently bifurcated the first quartile of the cost curve and structurally raised the mid-cycle incentive price — is exactly the kind of peak-narrative a price-taker tells at the top of a spread. Treat as hypothesis, not evidence (validated against history below).

Quality of earnings — normalize the one-times

The headline 2025 and Q1’26 numbers carry non-operating items that must be stripped before any run-rate or normalized-EPS work:

• FACT — Orica/Nelson Bros litigation gain (~$170M, Q1’26). The Q1 2026 print (net to common ~$615M, $3.98 dil EPS) includes a ~$170M pre-tax litigation settlement gain (cash received April 2026), per the Q1’26 call. This is a pure one-time and inflates the TTM-to-Q1’26 figures (EBITDA $3,493M, EPS $11.13 cont-ops basis). Normalized TTM EBITDA is closer to ~$3.3B. (Source: Q1 2026 earnings call, 2026-05-07.)
• FACT — 2025 had modest impairments, NOT a clean year. $25M asset impairment (Yazoo City AN asset group, following an incident that idled the plant) plus an electrolyzer-project abandonment charge. Immaterial vs. the franchise but cited by management as a drag on the 19% YoY net-earnings growth. (FY2025 10-K.)
• FACT — the historical impairment record is real and UK-concentrated: 2021 $521M ($285M goodwill + $236M long-lived/intangible, the UK Ince/Billingham assets), 2022 $239M (+$19M restructuring; further UK write-downs, Ince abandonment). These are the scars of the high-cost UK footprint, now largely rationalized (Ince closed 2022, ammonia at Billingham ceased 2023). (FY2021, FY2022 10-Ks.)
• FACT — unrealized gas-derivative MTM noise. 2025 ran a $5M unrealized MTM loss in COGS vs. a $35M gain in 2024 — a $40M non-cash swing that depressed 2025 reported gross margin vs. 2024. CF marks open gas derivatives to market through COGS; this is genuinely non-operating quarter to quarter. (FY2025 10-K.)

INTERPRETATION. Stripping the ~$170M Orica gain and the small impairments, 2025 was a clean, mid-cycle operating year — there is no GMPD-style hidden quality problem here (the QoE is high: no aggressive revenue recognition, no capitalized-cost games, FIFO gas accounting is conservative, derivatives marked through P&L). The earnings are real; they are simply cyclical.

Reconciling GAAP EPS $8.97 vs. ROIC’s $11.13 — the minority-interest treatment

This is a reconciliation the report must get right. The chain (FY2025 10-K consolidated statement of operations):

Total net earnings $1,798M → less net earnings attributable to noncontrolling interests $343M → net earnings attributable to common stockholders $1,455M → ÷ 162.2M diluted shares = GAAP diluted EPS $8.97.

FACT. The NCI is the ~11% economic interest CHS Inc. holds in CF Industries Nitrogen, LLC (CFN) — the consolidated subsidiary holding the bulk of CF’s US plants. CHS bought this strategic equity stake (2016, ~$2.8B) and in exchange receives ~11% of CFN’s earnings plus a urea/UAN supply offtake right. CF consolidates 100% of CFN’s revenue/EBITDA and then deducts CHS’s ~11% slice below the line as NCI. The CHS NCI sits on the balance sheet at $2,937M (2025) and is a real claim on cash flows — it must be added to EV (it is not common-shareholder value). NCI grew to $343M in 2025 (from $259M in 2024) simply because CFN earned more. (FY2025 10-K, Note 14; profile.)

INTERPRETATION — the ROIC $11.13 discrepancy. ROIC.ai’s TTM “EPS (cont-ops)” of $11.13 and the AZI valuation_index TTM EPS are computed on a pre-NCI or differently-normalized base (likely total net earnings, or TTM-to-Q1’26 which embeds the ~$170M Orica gain and a higher TTM net figure ÷ a lower live share count ~155M). The filing’s $8.97 is authoritative for GAAP FY2025. The ~$2/share gap is the combination of (a) the CHS NCI deduction, (b) the one-time Orica gain in the TTM window, and © share-count timing. The report should anchor on GAAP $8.97 (FY2025) and a normalized mid-cycle EPS materially below the TTM $11, and flag that aggregator P/E percentiles (AZI P/E 9.25x at the 18.6th percentile) are computed on near-peak/one-time-inflated earnings — the classic cyclical “cheapest P/E at the top” trap. (ROIC.ai, AZI accessed 2026-06-18; reconciled to 10-K.)

ROE/ROIC cyclicality

INTERPRETATION. A ROIC range of 6% (trough) → 41% (peak) over five years is the numerical fingerprint of a no-franchise commodity price-taker. The 2025 17% ROIC sits above CF’s own through-cycle average but below the 2022 spike. The capital base is the same gas-fed steel-and-pipe in every year; only the spread changes the numerator. By Greenwald’s test — does ROIC stay durably above WACC because of a barrier to entry? — CF passes only in the sense of a low-cost-position cost advantage (cheap North American gas + scale at Donaldsonville, the world’s largest ammonia complex), not a demand/captivity or network moat. The trough-year 6% ROIC is below WACC; the franchise does not generate excess returns through the cycle, it generates them when the spread is wide.

Balance sheet — genuinely conservative, the standout quality

This is where CF is unambiguously strong, and it is a deliberate, defensible management choice for a cyclical:

• FACT. Net debt/EBITDA 0.38x (2025), down from 2.2–2.8x in 2019–20; even at trough-2024 EBITDA it was 0.51x. Total debt/EBITDA 1.1x. (ROIC.ai credit ratios; reconciles to 10-K.)
• FACT. Total long-term debt $3,250M face / $3,215M carrying (2025), all now unsecured public senior notes (5.150% '34, 5.300% '35, 4.950% '43, 5.375% '44) after CF refinanced the $750M 4.500% senior secured notes due 2026 by issuing $1.0B 5.300% notes due 2035 in Nov 2025 (recognized a small $6M extinguishment loss). Moving from secured to unsecured is a credit-quality upgrade signal. No meaningful maturity wall before 2034. (FY2025 10-K, Debt note.)
• FACT. Cash $1,982M; net debt ~$1,233M. EBITDA/interest coverage ~22.6x (2025); EBITDA-less-capex/interest ~16x. Interest expense only $145M. (ROIC.ai; 10-K.)
• FACT. CF carries investment-grade ratings (Baa/BBB tier — agency letters not in the 10-K text; confirm exact notch from a rating-agency release. OPEN QUESTION). The conservative leverage and unsecured-only structure are consistent with low-BBB/Baa2-ish.

INTERPRETATION. A sub-0.5x net-leverage commodity producer with 22x coverage can fund growth capex, sustain buybacks, and survive a trough year without distress — exactly the right posture for a business with 2.4x EBITDA cyclicality. This is the most defensible single element of CF’s financial profile.

Cash conversion — the “industry-leading FCF” claim, examined

• FACT. 2025: CFO $2,752M, CapEx $950M → FCF ~$1,802M. CFO/EBITDA ~84%; FCF/EBITDA ~55%. 2024: CFO $2,271M, CapEx $518M → FCF ~$1.75B. 2023: CFO $2,757M, CapEx $1,724M (Waggaman) → FCF ~$1.03B. (FY2025 10-K cash flow statement.)
• INTERPRETATION. Management’s “industry-leading cash conversion” is largely fair on a maintenance basis: normalized sustaining capex is only ~$500–550M (the 2025 $950M and 2023 $1,724M are inflated by growth/M&A), depreciation ~$900M, low cash taxes in some years, and negative working-capital benefit from customer prepayments (forward sales). At ~$500M sustaining capex on ~$3.3B mid-cycle EBITDA, maintenance FCF conversion is genuinely high. Caveat: the figure is flattered in years where growth capex is excluded and will compress as Blue Point growth capex (~$400M/yr CF share) ramps 2026–2029 — reported FCF will look lower even though the business is healthy.

Verdict

Economics do NOT improve with scale in any franchise sense — this is a price-taker whose profitability is set by the natural-gas-to-nitrogen spread. The genuine, defensible edge is a supply-side cost advantage (cheap, abundant North American shale gas + the scale/logistics of Donaldsonville and the broader network), which delivers first-quartile cost position and outsized margins when the global cost curve is steep — but ROIC collapses toward/below WACC at the trough (6% in 2020). The earnings are high-quality (clean QoE, conservative FIFO/derivative accounting, no hidden problems) but low-durability (2.4x EBITDA cyclicality, ~$350M EBITDA per $1/MMBtu gas move). The balance sheet is a genuine strength — sub-0.5x net leverage, 22x coverage, no near-term maturities, unsecured-only — and is the right structure for a cyclical. Bottom line: a best-in-class operator of a structurally no-moat, cyclical, price-taking asset base, with a fortress balance sheet that lets it play offense at the trough.

7. Capital Allocation

CF’s capital-allocation record is the strongest part of the equity story — and the place where the Marathon capital-cycle lens raises the one watch-item.

The buyback engine — disciplined, counter-cyclical, and massive

FACT — repurchases (treasury-stock purchases, equity-statement / cash-flow basis, $M):

(Small differences between cash-flow “treasury stock purchased” and the equity-statement line are timing/settlement; both reconcile to the 10-K. FY2025 10-K.)

• FACT — share-count collapse. Diluted weighted shares: 193.8M (2023) → 180.7M (2024) → 162.2M (2025); against ~233M in 2017, that is a ~33% reduction over eight years. Live count ~155–158M. The 2024–25 alone retired ~18.5M + further shares. (FY2025 10-K; ROIC.)
• FACT — authorizations. A $3.0B program (authorized, effective through Dec-2025) was fully utilized (37.6M shares cumulatively repurchased); on May 6, 2025 the Board authorized a new $2.0B program through Dec-2029. ~$1.7B remained available at the Q1’26 call. (FY2025 10-K; Q1’26 call.)
• FACT — Q1’26 discipline tell. CF repurchased only $15M (150k shares) in Q1’26, explicitly because of geopolitical/conflict uncertainty, while stating shares trade “below intrinsic value.” (Q1’26 call, 2026-05-07.)

INTERPRETATION — did they buy well? Mostly yes, and counter-cyclically. The heaviest buying (2024, $1.5B) came at $69–94 — i.e., at/near the cyclical earnings trough, the correct time for a cyclical to shrink the float. The 2022 buying at $85–119 was at higher prices but funded by peak FCF, and even those shares are below the 2026 ATH of $137.60. The Q1’26 pause during the Iran/Hormuz spike — declining to chase a spiking stock — is a genuine discipline signal, not a red flag. This is textbook counter-cyclical capital return and the single best evidence for competent management. The ~33% share shrink has materially amplified per-share metrics through the cycle.

Dividends — modest, well-covered, room to grow

• FACT. Dividend held at $2.00/share in 2024–2025 (raised from $1.60 in 2023, +25%). Cash dividends paid $326M (2025), $364M (2024), $311M (2023). Payout ratio only ~15–22% of net earnings to common (and far less of FCF). (FY2025 10-K.)
• INTERPRETATION. CF deliberately keeps the dividend low and the buyback large — the right choice for a cyclical (a low fixed dividend is sustainable even at the trough; the variable buyback flexes with FCF). Plenty of headroom to raise, but management clearly prefers buybacks as the primary return vehicle.

M&A — Waggaman (Dec 2023): the one to scrutinize

• FACT. In December 2023 CF acquired the Waggaman, Louisiana ammonia facility for $1.675B (from Incitec Pivot/Dyno Nobel; ~880k tons/yr ammonia capacity, on the Mississippi River near Donaldsonville). It is now wholly owned (no CFN/CHS NCI). 2023 CapEx of $1,724M largely reflects this. (FY2025 10-K; 8-K Dec-2023.)
• INTERPRETATION — price/timing. At ~$1.675B for ~880k tons, the implied multiple is ~$1,900/ton of capacity — meaningfully below estimated greenfield replacement cost (>$3,000/ton for a new ammonia plant including multi-year lead time), and the deal closed at a cyclical low (2023, post-2022 peak, gas elevated). Buying brownfield capacity at well-below-replacement cost at a cyclical trough is good capital-cycle behavior (Marathon: buy assets when others won’t add capacity). The strategic logic — co-located on the same logistics corridor as Donaldsonville, immediately accretive ammonia tons, optionality for CCS/low-carbon — is sound. Verdict on Waggaman: well-timed, well-priced, on-strategy.
• FACT — historical M&A context. CF’s prior big deals: the transformational Terra Industries acquisition (2010, ~$4.7B) that created the scale leader, and GrowHow UK (2015, ~$580M for the remaining 50%) — the latter the source of the value-destructive UK exposure that produced the $521M (2021) + $239M (2022) Ince/Billingham impairments. So the M&A record is mixed across decades: domestically/scale-accretive deals have worked; the UK expansion destroyed capital. Waggaman returns to the winning playbook (low-cost North American gas).

Blue Point JV — the cycle-top growth bet (Marathon watch-item)

• FACT. CF is building Blue Point in Louisiana with JERA (Japan) and Mitsui as JV partners — an autothermal-reforming (ATR) + carbon-capture-and-sequestration (CCS) low-carbon ammonia complex adding ~1.5M tons/yr gross ammonia capacity, online late 2029, construction starting 2026 pending permits. CF’s share of capex is ~$400M/yr (2026 consolidated capex ~$1.3B, of which CF portion ~$950M = ~$550M sustaining + ~$400M Blue Point). JERA/Mitsui contributed $291M of NCI capital in 2025. (FY2025 10-K; Q1’26 call.)
• INTERPRETATION (Marathon lens). This is the item to watch. Adding ~1.5M tons of new ammonia supply, with first production in 2029, is a multi-year growth-capex commitment initiated near a cyclical-mid/high in the spread — precisely the behavior Marathon’s capital-cycle framework flags as value-destructive when an industry collectively adds capacity into strength. Three mitigants: (i) CF is sharing the capital and risk with JERA/Mitsui (de-risking, and securing a committed Japanese low-carbon ammonia offtake buyer), (ii) the project is explicitly low-carbon (45Q tax credits, CBAM premium into Europe, decarbonization premiums from Pepsi/POET-type buyers) — a differentiated product, not undifferentiated grey ammonia, (iii) brownfield/co-located economics. But the returns depend on a low-carbon-ammonia premium market that is still nascent, and 2029 supply lands into an unknowable spread. Verdict on Blue Point: defensible and de-risked via partners, but on probation — a cycle-aware investor watches whether this is the first move in an industry-wide capacity build that compresses the very spread CF depends on.

R&D and incentive alignment

• FACT. R&D is minimal (CF is a process manufacturer, not a technology developer); “innovation” spend is concentrated in CCS/decarbonization project capex, not income-statement R&D.
• FACT — incentive design (2026 proxy, FY2025 plan). Annual (short-term) incentive: 60% Adjusted EBITDA, 30% clean-energy/network-optimization milestones, 10% safety. Long-term incentive: 60% PRSUs / 40% RSUs; the PRSUs cliff-vest over 3 years on average RONA (Return on Net Assets) across three one-year periods, with a relative-TSR modifier (±20%). The committee explicitly declined to use a relative-TSR performance benchmark for the core metric, citing too few comparable-size peers. (DEF 14A 2026-03-17.)
• INTERPRETATION. There is no explicit ROIC/ROCE hurdle, which is a mild negative for a capital-intensive cyclical — but RONA is a genuine capital-efficiency metric (returns on the asset base, not just EBITDA growth), and the TSR modifier links pay to shareholder outcomes. The 60% EBITDA weight in the annual plan rewards the cyclical upswing management does not control (a common cyclical-comp flaw), but the LTI’s RONA focus partly offsets it. Net: reasonable, returns-aware comp design, short of best-practice (no hard ROIC gate, EBITDA-heavy STI).
• FACT — say-on-pay: ~94% support at the 2025 annual meeting. No shareholder revolt. (DEF 14A 2026.)
• FACT — insider ownership: de minimis. Every named director/officer is marked “*” (<1%); CEO Bohn 142,963 shares, ex-CEO Will 156,669. No founder/controlling holder; top holders are index funds (BlackRock 8.3%, State Street 5.1%). (DEF 14A 2026.)

Verdict

Management has allocated capital intelligently — this is the strongest pillar of the thesis. The buyback program is disciplined and counter-cyclical (heaviest at the 2024 trough, paused during the Q1’26 spike), has shrunk the float ~33%, and is the primary, FCF-flexed return vehicle alongside a deliberately modest, sustainable dividend. Waggaman (Dec-2023) was a well-timed, below-replacement-cost, on-strategy brownfield buy at a cyclical low. The balance sheet has been deleveraged to a fortress 0.4x. The one reservation is the Blue Point growth bet — a cycle-top, multi-year capacity addition that the Marathon lens flags, de-risked but not neutralized by the JERA/Mitsui partnership and the low-carbon differentiation. Incentive design is returns-aware (RONA-based LTI) but stops short of a hard ROIC hurdle and is EBITDA-heavy on the annual plan. Insider ownership is low and there are no open-market purchases — alignment runs through grants/RONA/TSR, not personal conviction buying.

7A. SEC Filings Sweep & Insider-Transaction Read

Material-event timeline (FY2021–Q2 2026) — built from the 8-K corpus

Insider-transaction read (251 Form 4 filings, ~Jul 2021 – May 2026)

I fetched and parsed all 251 Form 4 XMLs in the corpus. Transaction-code frequency: A (grants) 144 · F (tax-withholding) 128 · S (open-market sales) 121 · M (option/RSU exercise) 78 · G (gifts) 44 · I/D (other) 3.

• FACT — ZERO code-P open-market purchases in five years. Across all 251 filings there is not a single discretionary open-market purchase by any insider. Every acquisition is a grant (A) or option/RSU exercise (M); every disposition is a sale (S), tax-withholding (F), or gift (G). (Parsed from SEC Form 4 XML, CIK 0001324404, 2021-07 to 2026-05.)
• FACT — sell side. Approx. ~$296M of insider open-market sales (S) over the period, dominated by ex-CEO W. Anthony Will (~$172M), with Bert Frost (~$39M), Douglas Barnard (~$34M), CEO Christopher Bohn (~$17M), Richard Hoker (~$13M) the next-largest. Most sales cluster around RSU/PSU vesting events (the F tax-withholding code is the second-most-frequent, confirming a grant-vest-sell pattern); a portion of named-officer sales reference dispositions tied to vesting/diversification. Will’s large total is consistent with a retiring CEO monetizing a long-accumulated equity position.
• INTERPRETATION. The insider picture is neutral-to-mildly-cautionary but unsurprising for a large-cap with heavy equity comp. The total absence of open-market buying means there is no insider-conviction signal at any price — including the 2024 trough and the 2026 dip — but that is the norm for index-held mega-cap industrials, and CF insiders own little stock outright regardless. The selling is overwhelmingly mechanical (grant → vest → tax-withhold/sell), not a coordinated distribution signal. Read: no bullish insider tell; no alarming one either. The capital-conviction signal comes from the company’s counter-cyclical buyback, not from individuals’ wallets.

7B. Reconciliation Notes (ROIC/AZI vs. 10-K)

• EPS: Use GAAP diluted $8.97 (FY2025) per 10-K. ROIC/AZI TTM “$11.13 cont-ops” is pre-NCI / one-time-inflated (Orica) / lower-share-count and is not the GAAP figure — flag in valuation so percentile P/E (AZI 9.25x, 18.6th pctile) isn’t read as “cheap” on top-of-cycle + one-time earnings.
• Net earnings to common: $1,455M (2025) / $1,218M (2024) / $1,525M (2023) — 10-K matches log/ROIC.
• NCI: $343M (2025) = CHS ~11% of CFN; balance-sheet NCI $2,937M — confirmed in EV build.
• Debt: total debt $3,636M (incl. $311M operating-lease liabilities + $110M current); senior-note face $3,250M / carrying $3,215M — 10-K and ROIC reconcile.
• Credit rating: investment-grade Baa/BBB-tier inferred from structure; OPEN QUESTION — confirm exact agency notch (not disclosed in 10-K text).

8. Major Changes & Headwinds — Last Two Years

1. CEO transition: Tony Will → Christopher D. Bohn (2025). (FACT) Long-time CEO W. Anthony (Tony) Will, who ran CF from 2014 and architected the consolidation/buyback/Donaldsonville-CCS strategy, was succeeded by Christopher D. Bohn, the former CFO/COO — an internal, continuity-signaling handoff rather than a strategic break. (INTERPRETATION) Continuity is a double-edged sword: it preserves the disciplined capital-return culture but also the now-active “CF premium” / cycle-top-growth-capex posture; the CFO seat is currently held on an interim basis by Richard (Rich) Hoker (FACT), an open governance item until a permanent CFO is named.

2. Waggaman, LA ammonia complex acquisition (Dec-2023, ~$1.675B). (FACT) CF bought the ~880k-ton Waggaman ammonia plant from Incitec Pivot, lifting 2023 capex to $1,724M (vs a ~$500M sustaining norm). (INTERPRETATION) A counter-cyclical, low-multiple bolt-on of existing low-cost Gulf-Coast capacity — bought into a normalizing market rather than at the 2022 peak — and arguably the highest-quality capital-allocation move of the period: it adds tonnage at replacement-cost-favorable economics with no greenfield execution risk.

3. Blue Point JV with JERA + Mitsui (2025). (FACT) A Louisiana low-carbon ammonia project (autothermal reforming + carbon capture), +1.5M tons gross ammonia capacity, targeted online late 2029; construction starts 2026 pending permits. CF’s 2026 capex portion is ~$950M ($550M sustaining + ~$400M Blue Point); consolidated 2026 capex ~$1.3B. (INTERPRETATION) This is the cycle-top growth-capex flag (Marathon capital-cycle lens): a multi-year, greenfield, ~$4B+ gross-cost project commenced near a geopolitical earnings peak, with demand for low-carbon ammonia still nascent. The JV structure (JERA/Mitsui share cost and offtake) materially de-risks CF’s balance-sheet exposure, but execution, cost-overrun, and end-demand risk are real and back-end-loaded.

4. Decarbonization commercial traction. (FACT) Donaldsonville CCS is live; commercial low-carbon ammonia premium cargoes are flowing; named offtake/partnership progress with PepsiCo and POET; 45Q US tax credits and the EU CBAM premium are monetization levers. (INTERPRETATION) Genuine optionality, not yet a needle-mover — a call option on a carbon-priced world, appropriately treated as upside rather than base-case value.

5. Iran / Strait-of-Hormuz nitrogen supply shock (Q1’26). (FACT) Drove the March ATH (see the Five-Year Event Map above). (INTERPRETATION) The single largest thesis input of the period and the origin of the “CF premium” narrative (the relevant section) — but a transient supply event by nature, now partially unwinding on de-escalation.

6. Orica / Nelson Bros litigation gain (~$170M, Q1’26). (FACT) A one-time pre-tax gain (cash received April 2026) flattering Q1’26 net earnings (~$615M / $3.98 EPS). (INTERPRETATION) Non-operating — must be normalized out of run-rate earnings and FCF.

Verdict: net thesis-neutral-to-modestly-weakening at the margin. The Waggaman counter-cyclical buy and continued share shrink strengthen the long-run per-share story. But the cluster of cycle-top signals — a continuity CEO handoff with an interim CFO, a large greenfield growth-capex commitment (Blue Point), and an earnings/price peak built on a transient geopolitical premium plus a one-time litigation gain — collectively weaken the quality of the headline numbers the market is now anchoring to. The changes do not break the thesis; they raise the burden of proof on management’s “structurally-higher mid-cycle” claim.

9. Risk Analysis (Risk Matrix)

CF is, empirically and fundamentally, a commodity-price instrument (FactorsToday dominant factor = OilPrice β~1.2–1.32; market β only 0.165; idio vol 29.9%). The risk profile is overwhelmingly dominated by the nitrogen-price / natural-gas-spread cycle; most other risks are second-order amplifiers of that single variable.

Catastrophic-loss / total-loss risk: Low. CF owns irreplaceable, lowest-quartile-cost North American assets, carries minimal net leverage, and generates FCF even at trough nitrogen. The realistic downside is de-rating + earnings mean-reversion (a 30–50% drawdown, which the lifetime -76.7% max drawdown shows is within the historical envelope), not insolvency.

10. Valuation Discussion (Embedded Expectations)

No price target, no recommendation in this section — embedded-expectations and scenario framing only.

Spot snapshot (FACT, rebuilt to live EV)

At $102.93 (2026-06-18) on ~155–158M shares: market cap ~$16.0–16.3B; EV ~$20.2–20.5B (+ net debt $1.23B + CHS minority $2.94B). On TTM EBITDA of $3,493M that is EV/EBITDA ~5.8x and EV/sales ~2.8x. Headline GAAP P/E ~9.25x (TTM EPS ~$11.13 cont-ops basis) and FCF yield ~10–11% (~$1.65–1.8B FCF / ~$16B cap). P/B 2.98x, P/S 2.20x.

The cyclical-multiple inversion (the single most important valuation fact)

CF’s P/E is a contrarian indicator across the cycle (FACT, ROIC multiples + AZI valuation_index):

AZI own-history percentiles crystallize the peak-earnings mirage: P/E sits at the 18.6th percentile (looks cheap) against near-peak earnings, while the cycle-immune ratios are elevated — P/B 69.9th and P/S 68.8th percentile (composite 52.4th). (FACT, AZI 2026-06-18.) On the metrics that don’t flatter at the top of the cycle, CF is in the upper third of its own decade range — i.e., not cheap.

Embedded-expectations: what mid-cycle EBITDA does $102.93 imply?

The central question is what normalized EBITDA the price capitalizes. Bracketing through-cycle EBITDA (ASSUMPTION, anchored to the 2017–2025 actuals):

Embedded read (INTERPRETATION): at ~5.8x spot but ~6.3–7.2x on a $2.8–3.2B mid-cycle, the market is not pricing pure 2020-style trough reversion, nor the full 2022 peak as permanent. It is underwriting something close to “current elevated conditions persist as the new mid-cycle” — i.e., the price embeds a meaningful chunk of management’s “CF premium” thesis (a structurally higher mid-cycle nitrogen price driven by geopolitical bifurcation of the cost curve). If mid-cycle EBITDA is really ~$2.8–3.0B (the 2017–2021/2024 average zone), the stock is fairly-to-fully valued, not cheap. The “cheap P/E” is the trap.

Peer comparison (FACT, ROIC TTM EV/EBITDA + web, 2026)

CF’s ~5.8x sits below the diversified peers (NTR/MOS) and above Yara — consistent with a Deutsche Bank (Jan-2026) note flagging CF at a ~15% premium to Yara (FACT, web). The EV/sales premium (2.8x vs peers 0.9–1.8x) reflects CF’s superior margin structure (EBITDA margin ~46% vs Mosaic ~14%, Yara ~12%) and lowest-cost position — deservedly higher, but the absolute EV/EBITDA still trades through the cyclically-distressed European comp, which is the relevant cross-check on whether the “premium” is already in the price.

Scenario analysis (ASSUMPTION-driven; explicit nitrogen/gas inputs)

Valuation verdict

On P/E and FCF yield CF screens cheap; on EV/EBITDA, P/B, P/S and a normalized-EBITDA lens it is fair-to-full. The deciding variable is whether ~$3.5B TTM EBITDA is closer to the new mid-cycle (bull/“CF premium”) or a near-peak that mean-reverts toward ~$2.5–3.0B (bear/Marathon). The market is underwriting a partial “CF premium” — a mid-cycle structurally above the pre-2022 norm but below the 2026 geopolitical spike. That is the embedded bet a buyer at $102.93 is making.

11. Variant Perception

Consensus view. CF is the lowest-cost, lowest-risk nitrogen producer at the bottom of the global cost curve, with a fortress balance sheet (net-debt/EBITDA ~0.4x), a disciplined ~33%-since-2017 buyback, and — newly — a structurally higher mid-cycle nitrogen price because geopolitics (Iran/Hormuz, Russia, China/Egypt export curbs) has bifurcated the first-quartile cost curve into “low-cost-and-low-risk North America” vs “fragile, exposed” Middle East/Russia. The “CF premium” is the consensus’s organizing idea and the reason the stock holds ~5.8x EV/EBITDA / ~9x P/E with a 10%+ FCF yield rather than trading like a pure scrap-the-cycle commodity name.

Strongest bull case. (1) The capital cycle is genuinely tight: years of underinvestment plus the permanent loss/curtailment of marginal ME/Russian/European capacity raises the global incentive price, structurally lifting mid-cycle EBITDA toward $4–5B. (2) CF compounds per share through the trough via aggressive, value-accretive buybacks at a 10%+ FCF yield — the share count shrinks faster than the cycle decays. (3) Decarbonization (Donaldsonville CCS live, Blue Point ~95% decarbonized, 45Q + CBAM + Pepsi/POET offtake) is a free option on a carbon-priced world. (4) Net cash optionality + Waggaman shows counter-cyclical M&A discipline. If even half the “CF premium” is real, ~5.8x is cheap.

Strongest bear case. (1) This is a peak-earnings cyclical priced at a geopolitical-premium top. The “structurally higher mid-cycle” is the textbook narrative built at every commodity peak; the Marathon capital-cycle lens predicts high returns attract capital (incl. CF’s own Blue Point +1.5M t) and mean-revert. (2) The cheap P/E (18.6th pctile) is the classic peak-earnings mirage — the cycle-immune P/B (69.9th) and P/S (68.8th) say the stock is in the upper third of its own range. (3) The March ATH was a transient supply shock (Hormuz), already -25% unwound on US-Iran de-escalation + Henry Hub collapsing to ~$2.60. (4) China/Russia export policy can flood seaborne urea overnight; demand destruction caps the upside. (5) Growth capex (Blue Point) is being committed near peak earnings — the cycle-top capital-allocation error in miniature.

The 3–5 assumptions that matter most:

1. Is mid-cycle nitrogen structurally higher post-2022/2026? (The whole “CF premium.” If yes → bull; if it mean-reverts → bear.) — the master variable.
2. Does Chinese urea export policy stay restrictive? (The single largest swing factor in global N supply.)
3. What is normalized through-cycle EBITDA — ~$2.5–3.0B or ~$4–5B? (Sets whether ~5.8x is cheap or full.)
4. Does management keep buying back stock through the trough rather than over-spending on Blue Point at the peak?
5. Does North American natural gas stay cheap relative to global feedstock? (CF’s entire cost moat.)

Falsification tests.

• Bull case falsified if: NOLA/global urea reverts toward pre-2022 levels ($300–350/st) and holds for 2–4 quarters as Chinese exports resume — proving the “CF premium” was a risk-premium spike, not a structural shift; EBITDA settling at ~$2.0–2.4B would confirm.
• Bear case falsified if: nitrogen prices hold materially above the pre-2022 norm for multiple quarters despite a calm geopolitical backdrop and resumed Chinese exports — i.e., the higher incentive price proves demand/cost-curve-driven, not scarcity-driven; EBITDA durably ≥$3.5B without a supply shock would confirm.

Factor-positioning read (FactorsToday, 2026-06-18). CF is empirically an energy/commodity instrument: dominant loading is OilPrice β~1.2–1.32 (R² 0.30–0.40), Energy-sector β 0.67, and all factor-similar peers are oil & gas (OXY, SHEL, VLO, XOM, DVN, FANG, CHRD) — not ag-input names. Crucially, market β is only 0.165 (defensive to the broad index) with positive alpha (+0.134) and high idiosyncratic vol (29.9%) — the stock marches to the commodity/energy regime, not the S&P. The -25% pullback from the March ATH reads as a risk-premium unwind (rs_peak -24.89; m3 return deeply negative annualized, m6 still strongly positive; y3 +17.6%pa, y5 +16.8%pa) — i.e., the mean-reversion of a geopolitical spike inside a still-positive multi-year uptrend, not a structural falling knife (lifetime max drawdown -76.7% shows what a true cycle-break looks like — this isn’t that, yet). Where consensus may be offsides: the tape is pricing CF as a defensive, high-alpha energy proxy with a structurally re-rated floor; the bear’s variant is that the OilPrice/geopolitical beta cuts both ways and the same factor that drove the ATH is now reversing — the “CF premium” the market is capitalizing is, in factor terms, a geopolitical-risk premium that is actively deflating.

12. Fact vs. Interpretation Table

13. Open Questions

1. What is true normalized mid-cycle EBITDA? The whole valuation turns on whether it is ~$2.5–3.0B (bear) or ~$4–5B (bull/“CF premium”). Resolvable only with 4–8 quarters of post-shock data on nitrogen prices in a calm geopolitical backdrop.
2. Is the “CF premium” durable or a deflating risk premium? Will the higher incentive price survive resumed Chinese urea exports and Middle East capacity restoration? The tape is currently betting no.
3. What is the through-cycle cost of the CHS structure to common holders? CHS takes ~11% of CFN’s earnings (NCI $343M in 2025, growing pro-cyclically) plus a fixed offtake; the effective drag vs having retained 100% and financed Waggaman-style is unquantified.
4. Exact credit rating? Investment-grade Baa/BBB-tier inferred from the unsecured-only structure and 0.4x leverage; the precise agency notch is not in the 10-K text — confirm from a rating-agency release.
5. Does Blue Point mark the start of an industry-wide capacity build? If CF, the Gulf Coast greenfields (Ascension, St. Charles) and Middle East/FSU additions collectively land into the late-2020s, the very spread CF depends on compresses — the Marathon mechanism.
6. Will a permanent CFO be named, and does the Will→Bohn continuity preserve discipline or entrench the cycle-top-growth posture? Hoker holds the CFO seat on an interim basis.
7. Does the clean-ammonia energy market (marine fuel, Japanese/Korean co-firing) materialize, or stay subsidy-dependent vapor? It anchors much of the long-dated Blue Point/decarbonization narrative; independent analysis says it is heavily overstated to 2050.

14. What Must Be True

Bull case — what must be true

1. Mid-cycle nitrogen is structurally higher than the pre-2022 norm — the “CF premium” is real, driven by a permanently steeper/riskier cost curve, lifting normalized EBITDA toward ~$4–5B.
2. The global capital cycle stays disciplined — new low-cost capacity (Gulf Coast, Middle East, FSU, and CF’s own Blue Point) does not collectively flood the merchant pool and compress the spread.
3. North American gas stays cheap relative to global feedstock, preserving the first-quartile cost position.
4. Management keeps compounding per share — continued counter-cyclical buybacks at a 10%+ FCF yield shrink the float faster than the cycle decays; Blue Point earns its cost of capital.

Falsification test (bull): Global/NOLA urea reverts toward pre-2022 levels ($300–350/st) and holds for 2–4 quarters as Chinese exports resume, with EBITDA settling at ~$2.0–2.4B — proving the premium was a risk-premium spike, not a structural shift.

Bear case — what must be true

1. The 2025/2026 earnings are a cyclical/geopolitical peak that mean-reverts toward ~$2.5–3.0B EBITDA as the Hormuz premium unwinds and Chinese/Russian exports normalize.
2. The capital cycle bites — high returns attract the documented wave of new low-cost supply, compressing margins into the late 2020s (Marathon).
3. The “cheap” P/E is a peak-earnings mirage — the cycle-immune P/B (69.9th pctile) and P/S (68.8th) correctly say the stock is fair-to-full, and it de-rates as earnings fall.
4. Cycle-top growth capex (Blue Point) is value-dilutive — committed near peak earnings into a nascent low-carbon-ammonia demand market that disappoints.

Falsification test (bear): Nitrogen prices hold materially above the pre-2022 norm for multiple quarters despite a calm geopolitical backdrop and resumed Chinese exports, with EBITDA durably ≥$3.5B absent any supply shock — proving the higher incentive price is demand/cost-curve-driven, not scarcity-driven.

15. Source Appendix

See Appendix B — Source Appendix (appended to the combined report), which lists the primary and secondary sources relied upon throughout this memo.

APPENDIX A — Standard Diligence Questionnaire

CF Industries Holdings, Inc. (NYSE: CF) — Standard Diligence Questionnaire

Supplemental diligence appendix. Report date: 2026-06-19. Price referenced: $102.93 (2026-06-18 close). Fact / Interpretation / Assumption labels applied where it matters. This appendix carries no recommendation and no price target (the single position-taking exception lives in the Author’s Take block above).

General

What thoughtful questions have other investors asked about this company?

The debate among serious investors centers on one master question, with several sub-questions orbiting it:

• Is the “CF premium” real or a peak narrative? (INTERPRETATION) The single most-asked question is whether the post-2026 nitrogen price level represents a structurally higher mid-cycle (management’s claim that geopolitics has permanently bifurcated the first quartile of the cost curve into “low-cost-and-low-risk North America” vs “fragile, exposed” Middle East/Russia) or simply a deflating geopolitical risk premium. The stock answering this in real time: −25% off the March-2026 ATH of $137.60 as US-Iran tensions eased.
• What is true normalized mid-cycle EBITDA — ~$2.5–3.0B or ~$4–5B? (ASSUMPTION) This sets whether ~5.8x spot EV/EBITDA is cheap or full. The whole valuation turns on it.
• Is the cheap-looking P/E (18.6th own-history percentile) a value signal or the classic peak-earnings cyclical mirage? (FACT/INTERP) Sophisticated holders know cyclical P/E is a contrarian indicator (cheapest at the top); the cycle-immune P/B (69.9th) and P/S (68.8th) percentiles say the stock is in the upper third of its own decade range.
• Does Blue Point mark the start of an industry-wide capacity build that compresses the very spread CF depends on? (Marathon capital-cycle question.)
• Is management over-deploying growth capital at a cycle top while simultaneously claiming buyback discipline?
• What is the through-cycle cost of the CHS minority structure to common holders?

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? (FACT/INTERPRETATION) Near a cyclical high, not the trough. TTM EBITDA of ~$3.5B (which includes a one-time ~$170M Orica/Nelson Bros litigation gain; normalized ~$3.3B) sits well above the 2024 trough of $2.64B but below the 2022 energy-crisis peak of $6.42B. Q1’26 earnings were inflated by an acute Iran/Strait-of-Hormuz supply shock that spiked global nitrogen prices. The memo’s base case treats ~$2.8–3.2B as true mid-cycle, placing current earnings at the upper end of mid-cycle / lower end of peak — closer to a high than a low.

Driven by the external environment or internal actions? (FACT) Overwhelmingly external. Earnings are the arbitrage between the globally-set nitrogen price (driven by the marginal European/Asian gas-based producer) and CF’s own Henry Hub gas cost — both price-taker inputs CF does not control. EBITDA swung 2.4x ($6.4B → $2.6B) in two years on the spread alone, with flat ~19M-ton volume. The one genuine internal lever that compounds per-share value is the counter-cyclical buyback (~33% share-count reduction since 2017), but it moves per-share metrics, not the underlying earnings stream.

How stable are revenues? (FACT) Highly unstable in dollars, very stable in volume. Revenue arc ($M): 2020 4,124 / 2021 6,538 / 2022 11,186 / 2023 6,631 / 2024 5,936 / 2025 7,084 — a ~2.7x peak-to-trough swing. Volume is dead flat (19.1M / 18.9M / 19.1M product tons 2023–25). This is a price story, not a volume story. There is essentially no recurring revenue in the SaaS sense; ~all revenue re-prices each cycle to the global nitrogen benchmark.

Outlook for products/services? (FACT/INTERP) Nitrogen demand is anchored to food and grows only ~1–2%/yr (IFA). Near-term price outlook is the unwind of the 2026 geopolitical premium (Henry Hub back to ~$2.60 from a >$7 Feb settle; NOLA softening on import liquidation). The product mix is durable — granular urea and UAN (43–47% gross margins) outearn raw ammonia (31%) and explosives-grade AN (19%). Low-carbon ammonia (45Q + CBAM) is real but modest near-term optionality; clean-ammonia energy demand (marine fuel, Japanese co-firing) is largely subsidy-dependent and overstated on a 2030 horizon.

How big is this market — growing/shrinking, domestic/international? (FACT) Global, large, slow-growing. Global N-effective capacity +4% (166.3 Mt 2024 → 172.7 Mt 2026); global ammonia ~240 Mt (2023) → ~290 Mt by 2030. Demand grows ~1–2%/yr. CF is North American-centric in production (six US, two Canadian, one UK complex) but its prices are set in a globally-traded market with continuous import competition. US corn — the largest N-consuming crop — drove ~95.3M intended planted acres in 2026 (USDA NASS), a near-record but a marginal headwind if acreage shifts toward soybeans.

Business Quality & Competitive Moat

Is the industry getting more or less competitive? (FACT/INTERP) Roughly stable-to-slightly-more competitive on the supply side. Mild consolidation among merchant sellers (OCI’s exit/wind-down, Koch’s purchase of OCI’s Wever plant) reduces the seller count — a modest Marathon positive. But the offsetting force is larger: a documented wave of new low-cost Gulf Coast capacity (Ascension ~7.2 Mt ~2027, St. Charles ~3 Mt ~2027, CF’s own Blue Point ~1.5 Mt 2029) plus Qatar/Saudi/FSU additions is landing into elevated prices — the canonical late-cycle capital response that compresses margins.

How profitable is the business (ROIC, ROE)? (FACT) Spectacularly cyclical, not consistently elite. ROIC: 5.7% (2020, below WACC) / 18.2% (2021) / 41.3% (2022 peak) / 16.2% (2023) / 12.8% (2024) / 17.1% (2025). ROE 2025 ~37%. The trough ROIC dipping toward/below cost of capital is the tell: there is no durable franchise insulating returns from the cycle. Through-cycle ROIC is probably low-to-mid teens — adequate, entirely explained by cost-curve position, not a demand moat.

How profitable is the industry — competitors, barriers to entry? (FACT) Structurally a mixed-to-poor industry. CF’s own 10-K concedes ammonia/urea/UAN are “widely-traded fertilizer products and there are limited barriers to entry,” competing “based primarily on delivered price” — an unusually blunt self-admission of no moat. The industry is fragmented; the top-five merchant suppliers (Yara, CF, OCI [winding down], SABIC, Nutrien) hold only a limited share of global capacity. The single redeeming feature is real, persistent supply-side cost dispersion driven by regional gas geology, which hands first-quartile (US/Canada) producers a durable cost advantage over the European marginal price-setter.

Can the business be easily understood? (FACT) Yes — exceptionally clean. A pure-play converter of natural gas into ammonia and its derivatives via the century-old Haber–Bosch process, sold into agriculture (~75%+) and a higher-margin industrial tail (DEF, AN explosives, nitric acid). Profit = (globally-set nitrogen price − Henry Hub gas cost) × ~19M tons. No conglomerate noise. The clarity is a genuine analytical positive; the absence of pricing power is the defining structural fact.

Can it be undermined by foreign low-cost labor? (FACT/INTERP) Not by labor — by foreign low-cost feedstock. Nitrogen production is capital- and energy-intensive, not labor-intensive; labor is a trivial share of cost (gas is ~70–90% of cash cost). The real foreign threat is producers with cheaper gas: Middle East (Gulf gas, cheaper than US) and Russia (domestic gas, cheapest of all). CF’s defense is not cost-leadership on gas (it is not the cheapest) but low cost in a stable, CBAM-favored, logistically-integrated jurisdiction at the largest scale — trading a few dollars of gas cost for political stability and carbon-intensity advantage. Continuous import competition caps domestic prices (“we experience competition from foreign-sourced products continuously” — 10-K).

Do brands matter? (FACT) No. A ton of granular urea is a ton of granular urea. There is no brand pricing power; buyers (co-ops, distributors, traders, industrial users) purchase on delivered price.

Nature of competition? (FACT) Price-based, on delivered cost. Named competitors: Nutrien, Koch Fertilizer, LSB Industries, CVR Partners, Yara (10-K). Yara’s European fleet is the structurally cost-disadvantaged marginal producer that sets the price — the inverse of CF. Gulf/Russian producers are cheaper on gas but Hormuz-exposed / sanctioned / CBAM-penalized.

Customers’ switching costs? (FACT) Effectively zero. Ammonia/urea/UAN are fungible commodities to identical spec; switching to a rival molecule costs nothing. There are no contractual lock-ins of consequence beyond the CHS offtake right (which is CF’s own minority partner buying at market, not a captive third-party customer). The Greenwald verdict: a supply/cost advantage only — no customer captivity, no switching costs, no network effects, no brand — the weakest and most transient of the three genuine advantage types.

Financial Condition & Balance Sheet

Assets not fully recognized on the balance sheet? (INTERPRETATION) Yes, in the economic sense: CF’s irreplaceable low-cost asset base — Donaldsonville (the world’s largest ammonia complex), the broader six-US/two-Canada network, and the owned Corn Belt terminal/transport/logistics system — carries book value far below replacement cost. New ammonia capacity costs >$3,000/ton to build (multi-year lead time); CF bought Waggaman’s ~880k tons for ~$1,900/ton. The cost-curve position itself (cheap stranded US gas access) is an unbooked intangible. Conversely, there is no goodwill-style overstatement risk remaining after the UK write-downs.

Off-balance-sheet liabilities? (FACT/INTERP) Few of concern. The largest quasi-liability is on the balance sheet but is easy to miss as a common-holder claim: the CHS minority interest of $2,937M (CHS’s ~11% economic stake in CF Industries Nitrogen, LLC), which takes ~11% of CFN earnings (NCI $343M in 2025, growing pro-cyclically) plus a fixed urea/UAN offtake right — a permanent quasi-debt/quasi-equity claim on the core US earnings stream that must be added to enterprise value. Standard items: asset-retirement obligations, pension, operating-lease liabilities ($311M, on balance sheet under current GAAP), and a long-tail opioid-style legal exposure is not present here. The Orica/Nelson Bros litigation resolved in CF’s favor (~$170M gain).

How conservative is the accounting? (FACT/INTERP) Conservative and clean. No aggressive revenue recognition (commodity sales at delivery), FIFO gas accounting, gas derivatives marked to market through COGS (genuinely non-operating quarter to quarter — a $40M non-cash swing 2024→2025). The historical UK impairments ($521M in 2021, $239M in 2022) were taken promptly and the high-cost UK footprint rationalized (Ince closed 2022, Billingham ammonia ceased 2023). No hidden quality-of-earnings problem; the earnings are real, simply cyclical. The one thing an analyst must normalize is the ~$170M one-time Orica gain in the TTM window.

How CapEx-hungry? (FACT) Moderate maintenance, lumpy growth. Sustaining capex is only ~$500–550M/yr against ~$3.3B mid-cycle EBITDA — genuinely high maintenance-FCF conversion. But growth is capital-intensive and lumpy: 2023 capex hit $1,724M (Waggaman), and Blue Point growth capex (~$400M/yr CF share) ramps 2026–2029 (2026 CF-portion capex ~$950M = ~$550M sustaining + ~$400M Blue Point). Reported FCF will optically compress during the Blue Point build even though the business is healthy.

Capital Allocation & Management

How much FCF, how is it used, what is the philosophy? (FACT) 2025: CFO $2,752M − capex $950M = FCF ~$1,802M (FCF/EBITDA ~55%; maintenance FCF higher). Philosophy is textbook for a cyclical: a fortress balance sheet (net debt/EBITDA ~0.4x), a deliberately modest, well-covered dividend ($2.00/share, ~15–22% payout) as the fixed return, and a large, FCF-flexed buyback as the primary, variable return vehicle. This is the strongest pillar of the equity story.

Significant recent acquisitions? (FACT) Waggaman, LA ammonia facility (Dec-2023, ~$1.675B, ~880k tons/yr) — bought at ~$1,900/ton vs >$3,000/ton replacement cost, at a cyclical low, co-located on the Donaldsonville logistics corridor, now wholly owned. (INTERP) Well-timed, well-priced, on-strategy counter-cyclical M&A — the highest-quality capital move of the period. Historical record is mixed: Terra (2010, ~$4.7B, transformational/good) and the CHS/CFN transaction (2016) worked; GrowHow UK (2015) was the source of the value-destructive UK impairments. Waggaman returned to the winning playbook (low-cost North American gas).

Buying back shares? (FACT) Aggressively and counter-cyclically. ~33% share-count reduction since 2017 (233M → ~155–158M live). Buybacks ($M): 539 (2021) / ~1,370 (2022) / ~602 (2023) / ~1,535 (2024) / ~1,379 (2025). The heaviest buying (2024, $1.5B at $69–94) came at the cyclical earnings trough — correct timing. The $3.0B program was fully utilized (37.6M shares); a new $2.0B program runs through Dec-2029 (~$1.7B remaining). The Q1’26 discipline tell: only $15M (150k shares) repurchased during the Iran/Hormuz price spike, explicitly declining to chase the stock — genuine counter-cyclical discipline.

Issuing large amounts of new shares to insiders? (FACT) No. Equity comp exists (PRSU/RSU grants) but is modest relative to the ~33% net share reduction. Insider ownership is de minimis (every NEO/director <1%; CEO Bohn 142,963 shares, ex-CEO Will 156,669). Top holders are index funds (BlackRock 8.3%, State Street 5.1%); no founder/controlling holder.

Director/management compensation policy? (FACT/INTERP) Annual incentive: 60% Adjusted EBITDA / 30% clean-energy + network-optimization milestones / 10% safety. Long-term: 60% PRSUs (3-year cliff on average RONA across three one-year periods, with a relative-TSR ±20% modifier) / 40% RSUs. No explicit ROIC/ROCE hurdle (mild negative for a capital-intensive cyclical), and the 60% EBITDA-weighted STI rewards the cyclical upswing management does not control — but RONA is a genuine capital-efficiency metric and the TSR modifier links pay to shareholder outcomes. Say-on-pay ~94% support (2025). The committee declined a relative-TSR core benchmark (too few size-comparable peers). Net: returns-aware but short of best-practice.

Motivations of management? (INTERPRETATION) Continuity-driven and capital-returns-focused. The Will→Bohn (ex-CFO) succession in 2024 signals preservation of the disciplined consolidation/buyback culture rather than a strategic break — a double edge, since it also entrenches the cycle-top growth-capex (Blue Point) and “CF premium” posture. The CFO seat is held on an interim basis by Richard Hoker (an open governance item). The capital-conviction signal comes from the company’s counter-cyclical buyback, not individuals’ wallets — there have been zero open-market insider purchases in five years (251 Form 4 filings), with ~$296M of mechanical grant-vest-sell dispositions (ex-CEO Will ~$172M, a retiring-CEO monetization). No bullish insider tell; no alarming one either.

Valuation & Market Data

Is the stock an ADR, MLP, or K-1 issuer? (FACT) No — none of these. CF Industries Holdings, Inc. is a Delaware C-corporation, NYSE-listed common stock (ticker CF), issuing a standard Form 1099-DIV. Not an ADR (US domestic issuer), not an MLP, not a K-1 issuer. (Note: the related but distinct CVR Partners is an MLP nitrogen play — CF is not.)

Dividend policy? (FACT) Modest and deliberately conservative: $2.00/share annual (raised from $1.60 in 2023, +25%; held flat 2024–2025), ~15–22% payout of net earnings to common, ~$326M paid in 2025. Yield ~1.9% at $102.93. Plenty of headroom to raise, but management clearly prefers buybacks as the primary return vehicle — the correct structure for a cyclical (a low fixed dividend survives the trough; the variable buyback flexes with FCF).

How profitable? (FACT) Very, at this point in the cycle: FY2025 gross margin 38.5%, EBITDA margin ~46%, operating margin ~33.5%, ROE ~37%, ROIC ~17%. But none of these are structural — they are a snapshot of where the gas-nitrogen spread sat in 2025. At the 2020 trough, gross margin was 19.4% and ROIC 5.7%.

Is net income diverging from cash from operations? (FACT) No material divergence — cash conversion is strong and clean. 2025 CFO $2,752M vs total net earnings $1,798M (CFO/net earnings ~1.5x, helped by ~$900M D&A and working-capital benefit from customer prepayments). CFO/EBITDA ~84%. This is a cash-generative, not an accrual-flattered, business — the opposite of a quality-of-earnings red flag. The only normalization needed is the ~$170M one-time Orica litigation gain in the TTM net-income figure.

Risks & Downside

What factors would cause the stock to decline? (FACT/INTERP) In order of importance: (1) gas-nitrogen-spread mean-reversion — the dominant risk; a single spread variable swings EBITDA ~4x; (2) geopolitical-premium unwind — already underway (−25% off ATH on US-Iran de-escalation); (3) Chinese/Russian export-policy normalization — a resumption of Chinese urea exports would flood seaborne supply and compress NOLA; (4) new global capacity landing into elevated prices (Marathon mean-reversion, including CF’s own Blue Point); (5) demand destruction at high prices (farmer affordability caps urea); (6) Blue Point execution/cost overrun; (7) a US natural-gas spike without a matching nitrogen-price move (CF is unhedged forward). The realistic downside is a de-rating + earnings mean-reversion drawdown of perhaps 30–50% — within the historical envelope (lifetime max drawdown −76.7%).

Risk of a catastrophic loss? (INTERPRETATION) Low. CF owns irreplaceable, lowest-quartile-cost North American assets, carries minimal net leverage (net debt/EBITDA ~0.4x, ~22x interest coverage, no maturity wall before 2034), and generates FCF even at trough nitrogen prices. The fortress balance sheet is precisely the posture a 2.4x-EBITDA-cyclical needs to survive a trough without distress.

Chance of a total loss? (INTERPRETATION) Negligible. This is not a balance-sheet-impaired or technologically-obsolescent business. The Haber–Bosch chemistry and the food-demand base are permanent; the assets are long-lived and low-cost. Total-loss scenarios would require a combination of catastrophe (multi-plant destruction) and balance-sheet impairment that the conservative leverage makes implausible. The bear case is lower returns and a de-rating, not insolvency.

Recent News & Events

Has the business environment changed recently? (FACT) Yes — acutely and transiently. A late-Q1’26 Iran conflict and Strait-of-Hormuz closure delivered the third major nitrogen supply shock in six years (management cited 31 ME ammonia plants impacted, 49 South-Asian curtailed on LNG feedstock, ~20 Russian plants drone-hit, Egypt’s +$90/t export duty, China restricting exports). Global urea spiked (a decade-record weekly move); CF hit an ATH of $137.60 (Mar-2026). The premium has since partly unwound (US-Iran de-escalation; Henry Hub back to ~$2.60), with the stock at $102.93. The structural question is whether this reset the mid-cycle higher (management’s “CF premium”) or was a deflating risk premium (the bear/Marathon view).

Significant acquisitions? (FACT) Waggaman (Dec-2023, $1.675B) is the most recent completed acquisition. The current major capital project is the Blue Point JV (announced/advanced 2025; CF 40% / JERA 35% / Mitsui 25%) — a Louisiana low-carbon ammonia complex (ATR + CCS, ~95% CO₂ capture), +1.5M tons gross capacity, online late 2029, construction starting 2026. This is the Marathon cycle-top growth-capex watch-item, de-risked but not neutralized by partner cost/offtake sharing.

Accounting-policy changes? (FACT) None material. The notable financial-structure change was the Nov-2025 refinancing: issued $1.0B 5.300% unsecured notes due 2035 and repaid the $750M 4.500% senior secured notes due 2026 — a secured→unsecured shift that is a credit-quality upgrade signal (small $6M extinguishment loss).

Recent changes — new markets, facilities, management? (FACT) (1) Management: CEO transition Tony Will → Christopher D. Bohn (2024); interim CFO Richard Hoker (open item). (2) Facilities: Donaldsonville CCS live (July 2025); Verdigris nitric-acid abatement project (Q4’25, cut CO₂e >600,000 t/yr); Yazoo City CCS targeted ~2028; UK operations restructured ($23M charge, 2025); Blue Point under construction. (3) Markets: first low-carbon ammonia premium sales into Europe/Africa (2025); CBAM definitive phase began 1-Jan-2026 (certificate purchases ~Feb-2027), advantaging low-carbon North American exporters; 45Q ($85/mt CO₂) monetization on Donaldsonville CCS. (4) One-time: the ~$170M Orica/Nelson Bros litigation settlement gain (cash received April 2026).

Where a question does not map cleanly to a commodity nitrogen producer (e.g., recurring/subscription revenue, brand value, switching costs, ADR/MLP status), the answer states so explicitly and gives the correct sector analog. Primary sources are catalogued in the Source Appendix.

APPENDIX B — Source Appendix

CF Industries Holdings, Inc. (NYSE: CF) — report date 2026-06-19. Sources relied upon throughout this note, categorized below. Primary (regulatory filings, company disclosures) prioritized over secondary; all third-party data is reconciled to primary filings where a primary source exists.

Primary Sources — SEC Filings & Company Disclosures

Quantitative Data Sources

Secondary & Industry Sources

Authority/reconciliation note: SEC filings (items 1–10) and CF’s own disclosures are primary and authoritative. ROIC.ai, AZI, and FactorsToday (items 11–14) are third-party/internal aggregated data used for cross-checks and the factor/price-positioning overlay; every material figure was reconciled to the underlying filing, and discrepancies (e.g., ROIC/AZI TTM “$11.13” EPS vs GAAP $8.97) are flagged in the memo. Secondary and industry sources (items 15–27) inform structural, regulatory, and peer context but do not override primary financial figures.