Digital Realty Trust, Inc. (NYSE: DLR) — The Operating Bull Won; the Capital-Light Test Did Not

Independent research memo. Report date: 2026-08-20; market data through the 2026-08-19 close. This is an update to the 2026-06-20 report. The analytical body carries no recommendation and no price target; the sole exception is the clearly labeled Claude’s Take below.

⚡ Claude’s Take

This is the author’s independent opinion, provided for general information. It is not investment advice.

Verdict: HOLD / avoid adding here; accumulate on a genuine de-rate; not a short. Digital Realty’s operating quarter was better than the June report expected. At its share, it produced $208 million of Q2 bookings, a $1.4 billion backlog before two large July leases, and 11.5% expected stabilized development yields. Cash renewal spreads reached 25.4% in Q2 and 11.0% over the trailing year. Core FFO per share excluding the promote rose 14%, and the 2026 guide rose to $8.15–$8.20. The bear case that pricing would fade immediately has been falsified.

The price still offers no margin of safety. At $192.35, DLR trades at 23.5x the 2026 Core FFO guide, approximately 25.4x a clean two-quarter AFFO annualization sensitivity, a 2.54% dividend yield, and the 97th percentile of its own price/book and price/sales history. Its enterprise-value and distributable-cash multiples are now essentially the same as Equinix’s, even though Equinix has more than twice the physical interconnections, one-fifth the largest-customer concentration, and materially better demonstrated asset economics. My accumulation zone is approximately $150–170. I would become genuinely interested at the lower end or after evidence that DLR can compound per-share FFO while keeping annual share growth below 3%.

The updated framing is “the operating bull won; the capital-light test did not.” Q2 Core FFO of $2.65 included a $188 million promote, and another $27 million came from a business-interruption insurance recovery. Removing both leaves about $2.06 per share, still approximately 10% above last year. Clean AFFO was about $1.87, approximately 11% higher. Those are good numbers. But shares and units outstanding grew 7.7% in six months, DLR raised $2.5 billion through the ATM, issued 12.3 million shares for the Blackstone assets, and committed more equity to Teraco and Columbia Capital. Private capital is supplementing common equity, not replacing it. The strict June bull-break test — dilution above 5% while FFO per share stalls — has not triggered because per-share growth remains strong. The cleaner confirmation test — 8–10% per-share growth with share growth below 3% — failed decisively. Conviction: medium-high. What flips me bullish is a de-rate or two years in which realized development returns remain double-digit, clean FFO per share compounds, and diluted share growth falls below 3%. What flips me bearish is new-start yields below 9–10%, renewal spreads returning toward flat, or another high-single-digit expansion in economic units without a durable per-share payoff. Tag: the data-center engine is running beautifully; the share printer is running too.

What changed since 2026-06-20

📈 Stock Price Action — Five-Year Event Map

Price moves are facts; attributed causes are interpretations. Adjusted closes are used.

DLR fell from roughly $150 in late 2021 to $77.88 on 2023-05-24, then re-rated with AI demand, improving spreads, falling leverage, and the private-capital narrative. It reached a record $202.56 on 2026-04-20 and closed at $192.35 on 2026-08-19, up 19.9% over twelve months and 45.8% over five years, but only 5.0% below the all-time high.

The stock is statistically still a REIT, not a semiconductor proxy. FactorsToday assigns loadings of Real Estate +1.06, Market +0.86, Interest Rate −0.47, and Momentum +0.07, with R² of 0.53. Its closest factor neighbors are real-estate ETFs and infrastructure REITs. AI demand is fundamental; rates remain the largest market transmission channel.

1. Executive Summary

Digital Realty is a global data-center REIT operating two businesses under one roof. The higher-quality business is retail colocation and interconnection: many customers, dense network ecosystems, meaningful switching costs, and recurring cross-connect revenue. The larger business is wholesale/hyperscale: long leases for large blocks of powered capacity, attractive during scarcity but structurally more reproducible and concentrated. The consolidated moat is therefore narrow and asymmetric.

Q2 strengthened the operating case. DLR had 310 data centers, 235,500 cross-connects, more than 6,000 customers, 3,102 MW of installed IT load, and 2,799 MW occupied. At DLR’s share, the quarterly booking mix was $87.8 million of 0–1 MW, $20.5 million of interconnection, and $99.8 million of greater-than-1 MW annualized GAAP rent. After quarter-end, two hyperscale leases added $205 million at DLR’s share. Backlog reached $1.4 billion at DLR share before those July leases.

The weakness is ownership economics. DLR grew revenue from approximately $3.2 billion in 2019 to $6.1 billion in 2025, yet Core FFO per share was nearly flat from 2021 through 2024. Share issuance and purchased scale absorbed the growth. The recent per-share inflection is real — $7.39 in 2025 and $8.15–$8.20 guided for 2026 — but the funding pattern has not changed as much as the “capital-light” label implies.

Valuation capitalizes success. DLR now trades at approximately the same forward distributable-cash and enterprise-value multiples as Equinix. The operating evidence supports continued growth; the multiple assumes that growth accrues to each share, development yields survive the capital cycle, and the current scarcity regime lasts. The report’s central question is no longer whether demand exists. It is whether management can convert exceptional demand into durable per-share value without repeating the dilution-heavy history.

2. Business Profile and Revenue Model

DLR sells powered data-center space, cooling, physical security, and connectivity through PlatformDIGITAL. The product spectrum ranges from cabinets and small deployments to entire hyperscale buildings. Cross Connect, Metro Connect, ServiceFabric, Internet access, pathway services, and cloud on-ramps create the connective layer. A newer business manages private-capital vehicles and earns management, development, construction, leasing, fit-out, and performance fees.

The portfolio is global, but operating quality differs by market. At June 30, Americas occupancy was 95.7%, including Northern Virginia at 98.6%, Chicago at 97.1%, and Dallas at 93.9%. EMEA occupancy was 82.1%; London was 71.9%, Amsterdam 79.7%, and Frankfurt 93.3%. APAC was 84.1%, with Singapore at 93.9%. This dispersion is important: “power scarcity” is not one global market condition.

Revenue is contractual and recurring, but concentration is material. The largest customer — described as a Fortune 50 software company — represented 10.8% of annualized recurring revenue; Oracle represented 10.0%. The top two were therefore 20.8%, and the top 20 were 53.1%. Credit quality is generally strong, and management says the large leasing cohort is traditional investment-grade hyperscalers rather than speculative neoclouds. Still, a few sophisticated customers retain bargaining power and can self-build.

The product-mix evidence supports the two-speed thesis. Greater-than-1 MW leases represented 62.5% of annualized data-center rent, compared with 33.0% for 0–1 MW. Physical cross-connects grew 2.8% year over year while occupied IT capacity grew 9.1%. Same-capital interconnection and other revenue grew 9.0%, a positive signal, but leased power is expanding faster than the network graph.

Verdict: a stable, global recurring-revenue platform with a growing fee layer, but one whose majority economics remain hyperscale land-and-power rather than network-dense interconnection.

3. Industry Structure and Capital Cycle

Data centers are locally constrained, not universally scarce. CBRE’s Q1-2026 survey showed year-over-year inventory growth of 33% in the four largest North American markets while vacancy remained 0.3% in Northern Virginia, 1.0% in Atlanta, 1.8% in Dallas-Fort Worth, and 2.2% in Chicago. Absorption matched new supply in Northern Virginia. That is extraordinary near-term landlord leverage.

The counterexamples matter equally. European vacancy was 5.0% in Frankfurt, 8.6% in London, 9.4% in Amsterdam, and 6.7% in Paris. Querétaro vacancy rose from 0.9% to 10.6% as new hyperscale capacity outran demand. Chicago rents rose 14.7%, while Atlanta rose only 2% and Dallas was roughly flat. Grid queues and permitting create scarcity; when power arrives, local economics can normalize quickly.

The Greenwald framework separates genuine captivity from supply scarcity. Interconnection hubs benefit from customer captivity and local network effects: each incremental participant makes the ecosystem more useful, and moving a dense deployment is operationally painful. Wholesale facilities benefit from scale, execution, and secured power, but their product can be reproduced by other well-funded developers or hyperscalers themselves. DLR’s current advantage in hyperscale is a valuable supply position, not permanent customer captivity.

The Marathon capital-cycle warning is intensifying. DLR had $20.18 billion and 1,402 MW under construction, up 34% sequentially, plus roughly 9 GW of future runway. H1 real-estate improvements were 1.77x depreciation and amortization. Equinix raised its 2027–2029 plan to $5–7 billion of annual capital expenditure. Industry inventory across CBRE’s tracked markets expanded 25%, and private capital continues to target the sector. Power delays postpone the clearing event; they do not repeal the supply response.

Verdict: excellent near-term fundamentals inside a late capital cycle. The scarcity bull and supply bear are both right, but on different time horizons.

4. Competitive Position and Moat

DLR’s moat is narrow, local, and asymmetric.

1. Retail colocation and interconnection — real customer captivity. Record 0–1 MW plus interconnection bookings, 142 new logos in Q2, 9% same-capital interconnection growth, and 235,500 physical connections show a real ecosystem. Connected campuses, cloud on-ramps, and ServiceFabric increase switching costs.
2. Hyperscale wholesale — scale without captivity. DLR can secure power, entitle land, finance construction, and deliver globally. Those are meaningful advantages. They are partly purchasable by capital and do not prevent a hyperscaler from using another provider or self-building at renewal.
3. Private-capital platform — potentially capital-efficient, not yet proven. Managing third-party capital can add fee income and reduce balance-sheet intensity. The simultaneous ATM issuance, land acquisition, minority buyouts, and Blackstone asset purchase show that the model is not yet truly asset-light.

The Equinix comparison remains decisive. EQIX reported approximately 522,700 interconnections versus DLR’s 235,500, and 2.3% largest-customer concentration versus DLR’s 10.8%. Equinix’s 27% stabilized cash-on-cash return is backward-looking on mature assets; DLR’s 11.5% is a projected yield on current construction, so the figures are not directly comparable. Even with that caveat, DLR has not demonstrated Equinix-like asset economics.

Expected development yield is the best current quantitative test of DLR’s supply advantage. At 11.5%, the pipeline still clears current borrowing costs and the June report’s 9–10% danger threshold. Regional mix deserves scrutiny: Americas was expected at 11.8%, APAC 11.2%, and EMEA 10.4%. EMEA construction was only 12% leased at quarter-end, versus 62% in the Americas. The aggregate pipeline was 54% leased at June 30 and approximately 63% after July signings.

Verdict: DLR is the scale leader in a weaker profit pool and the clear number two in the stronger one. Current power scarcity enhances its returns but does not broaden the moat.

5. Growth History and Forward Opportunities

The historical record is growth without adequate per-share translation. Core FFO per share was approximately $6.61 in 2021, $6.70 in 2022, $6.55 in 2023, and $6.72 in 2024. It rose 10% to $7.39 in 2025 and is guided to $8.15–$8.20 in 2026. The inflection is now supported by two years of evidence, but not yet by restrained issuance.

Forward visibility is strong:

• Backlog: $1.9 billion at 100% share and $1.4 billion at DLR share before July leases, equal to roughly 30% of in-place rent. Management identified approximately $635 million of commencements in H2-2026, $480 million in 2027, and $312 million in 2028 and later.
• Bookings: $307 million at 100% share in Q2; record $108 million in 0–1 MW plus interconnection. July added $410 million at 100% share from two hyperscale leases.
• Renewal mark-to-market: Q2 cash spreads were 5.2% for 0–1 MW and 66.7% for greater-than-1 MW, with the wholesale number driven by a few Singapore leases. The better normalized figures are 11.0% trailing-twelve-month blended spreads and the 9–11% full-year guide.
• Occupancy: same-capital occupancy rose 100 basis points year over year to 92.5%; full-year improvement guidance rose to 75–125 basis points.
• Development: $20.18 billion of expected investment at 100% share, 1,402 MW under construction, 11.5% projected yield, and a large secured-power runway.
• Fees: excluding the promote, quarterly fees were a little above $45 million. Columbia adds private-capital relationships and more than $9 billion of commitments, while management says $10–12 billion can be deployed over one to two years.

There are three quality caveats. First, the July bookings are mostly hyperscale, not the higher-captivity retail layer. Second, part of the rent-guide raise is non-cash: expected non-cash rent adjustments worsened by $55 million to negative $145–150 million. Third, development and land-banking require external funding years before stabilization.

Verdict: unusually strong contracted growth with good near-term visibility. The unknown is not demand; it is the per-share return after funding and the durability of current yields when supply energizes.

6. Financial Quality and Quality of Earnings

Q2 and H1 bridge

The promote was economically earned through successful development and leasing; it is not fake revenue. It is still non-recurring. DLR recognized $201 million of gross promote revenue, approximately $14 million of current expense, and $188 million of net Core FFO benefit. Another $35 million of related compensation is expected over two years. The correct recurring comparison excludes the promote and the $27 million insurance recovery.

AFFO is useful but optimistic. Q2 recurring capital expenditure was $76.7 million and H1 was $136.3 million, while full-year recurring capex plus capitalized leasing-cost guidance remained $400–425 million. AFFO also adds back $29.3 million of internal leasing compensation in H1. A quarter or half-year run rate therefore cannot be annualized without allowing for back-half maintenance spending.

True post-development free cash flow is negative. H1 operating cash flow was $1.60 billion, compared with $1.78 billion of real-estate improvements, $2.17 billion of acquisitions, and approximately $0.41 billion of net equity-method contributions. The cash deficit before dispositions and financing was approximately $2.77 billion and was funded largely with $2.50 billion of common-equity proceeds. ROIC.ai’s “free cash flow” field should not be used because it does not reliably subtract DLR’s development capex.

Same-capital operations were strong. Q2 cash NOI grew 8.9% reported and 7.2% constant currency; H1 grew 8.4% reported and 4.9% constant currency. This gap shows that FX helped reported growth. It does not negate the improvement, but constant currency is the better baseline.

The balance sheet is sound. Consolidated debt was $18.64 billion; management’s pro-rata calculation produced $18.54 billion of net debt, 4.7x net debt/LQA adjusted EBITDA, and 5.2x fixed-charge coverage. Debt was 92% unsecured and 90% fixed at a 2.9% average coupon with a 4.5-year average maturity. The issue is the forward cost: expected long-term debt pricing rose to 4.5–5.5%, while development spending accelerated.

Equity-method accounting obscures real cash economics. Unconsolidated ventures generated $301.6 million of Q2 NOI at 100%, while DLR’s share was $102.8 million of NOI, $93.9 million of cash NOI, and $58.1 million of Core FFO. GAAP equity-method income was nearly zero after depreciation and other accounting. For this REIT, per-share FFO, cash NOI, development returns, and economic-unit growth are more informative than GAAP earnings.

Verdict: clean operating growth is approximately 10%, not the 40% implied by headline Core FFO. That is still excellent. Financial quality is constrained by aggressive external funding, optimistic AFFO maintenance assumptions, and a growing web of JVs and transactions.

7. Capital Allocation and Incentives

The quarter is best understood as a capital-allocation bundle, not a single transaction.

The Blackstone trade can create value, but the proof is incomplete. DLR issued shares at an approximately 4.29% forward Core FFO yield and 2.56% dividend cost to acquire assets advertised at above a 6.5% stabilized cap rate with 3.6% escalators. That spread is favorable. Yet the cap rate is on completed gross asset value, while the assets require more funding and do not fully stabilize until 2027–2028. DLR did not disclose the incremental NOI, interest, remaining capex, or per-share accretion bridge.

The dilution language requires precision. Shares and units rose from 349.746 million at year-end to 376.675 million at June 30, +7.7%. Core-FFO/AFFO weighted diluted shares rose 3.7% in H1 and 4.9% in Q2 because issuance occurred during the period. Pro-forma outstanding growth including transaction stock is greater than 7% from the March proxy baseline and approximately 9.3% from year-end, but that is not identical to EPS dilution because acquired Teraco earnings accompany the shares. DLR assumed the put was share-settled in GAAP EPS/Nareit FFO, but explicitly excluded those potential shares from Core FFO/AFFO before settlement. The right conclusion is that common-equity dependence accelerated, while clean per-share FFO still grew fast enough to absorb it — so far.

Incentives are directionally aligned but incomplete. The CEO annual plan weighted 45% to Core FFO/share, 15% to same-store cash NOI, 15% to small-footprint/interconnection signings, and 25% to individual goals. Long-term awards use relative TSR and same-store cash NOI. Those are better than raw revenue or asset-growth goals. There is no ROIC, development-yield, or dilution hurdle. The carried-interest plan can allocate up to 50% of promote distributions to employees, aligning staff with partner returns but also rewarding transaction velocity.

The insider record needs a correction from June. Former CEO A. William Stein made two genuine open-market purchases in 2022. There have been no code-P purchases since September 2022 and none under Andrew Power. Current executives and directors own less than 1% in aggregate. The latest July filings were grants and tax withholding, not discretionary buying or selling.

Verdict: plausible value creation financed with expensive but richly valued equity, alongside a renewed asset-growth surge. The capital-light platform is strategically useful; it has not displaced the ATM or on-balance-sheet growth.

8. Changes and Headwinds

The most important changes since the prior report are:

• Pricing accelerated. The full-year renewal-spread guide rose 250 basis points to 9–11%. The extreme 66.7% greater-than-1 MW Q2 spread is episodic, but 11% LTM is not.
• The pipeline expanded. Under-construction capacity increased to 1,402 MW and expected investment to $20.18 billion while the projected yield held at 11.5%.
• Guidance rose. Revenue ex-promote increased to $6.85–6.95 billion, adjusted EBITDA to $3.75–3.85 billion, and ex-promote Core FFO/share to $8.15–8.20.
• Capital intensity rose faster. Net development-capex guidance increased $750 million to $4.25–4.75 billion; expected debt pricing worsened.
• The private-capital story became more complex. Columbia deepens the manager, while the Blackstone buyout reconsolidates assets and Kansas City expands land banking.
• Equity dependence returned visibly. The ATM raised $2.5 billion in H1 and outstanding economic units rose 7.7% in six months.
• Regulatory scrutiny is moving toward power costs. DLR’s August Texas release explicitly addressed grid reliability, consumer protection, and responsible resource management. This is not currently a financial impairment, but it makes utility cost allocation, curtailment, and permitting more important.

The near-term tape is stronger; the long-term supply setup is worse. That combination is typical near the later part of a capital cycle.

9. Risk Analysis

Catastrophic loss risk remains low. DLR owns hard assets, has investment-grade access, mostly fixed unsecured debt, diversified geography, and 4.7x leverage. The plausible downside is a multi-year multiple de-rate and weak per-share returns, not insolvency.

10. Valuation Discussion — Embedded Expectations

At the 2026-08-19 close of $192.35:

The own-history comparison is stronger than the cross-sectional one. DLR’s 97th-percentile P/B and P/S show that investors already capitalize the AI/power-scarcity regime as durable. AMT and CCI have not received comparable re-ratings; IRM has. The sector is not uniformly expensive, so DLR’s premium is specific to data-center growth.

The Equinix discount has disappeared on the metrics that matter. EQIX at $1,077 trades around 25.1x its approximately $43 FY2026 AFFO guide and roughly 24.6x adjusted EBITDA. DLR trades around 25.4x clean AFFO and 24.5x guided EBITDA. Equal multiples are difficult to justify when EQIX has the stronger network graph, lower concentration, and better demonstrated returns. DLR’s higher 2.5% dividend yield versus EQIX’s roughly 1.9% is not enough compensation because DLR’s payout is frozen and its funding need is larger relative to retained cash.

Reverse the multiple using the $7.58 clean AFFO sensitivity and flat $4.88 annual dividends. To earn an 8% five-year total return, clean AFFO/share must compound approximately 11.2% at a 20x exit, 9.1% at 22x, 7.3% at 24x, or 5.6% at 26x. A normal 20–22x exit therefore requires roughly 9–11% clean per-share growth for five years. The current price does not leave room for ordinary growth plus an ordinary de-rate.

Scenario ranges — not price targets

The ranges exclude dividends and are mechanical outcomes, not forecasts. Including three years of dividends produces only approximately 4–7% annualized total return in the base case because the current price already capitalizes much of the operating success. Capital structure choices matter almost as much as leasing.

Verdict: excellent assets at a near-peak multiple, with essentially no relative discount to the better franchise. Embedded expectations require both strong operations and substantially better per-share capital discipline.

11. Variant Perception

Consensus: DLR is one of the cleanest public ways to own AI infrastructure. Power scarcity is durable, legacy wholesale leases are below market, backlog makes 2027–2028 growth visible, leverage is repaired, and private capital converts a balance-sheet-heavy REIT into a fee-earning asset manager.

Strongest bull case: renewal pricing is not merely cyclical because utility, entitlement, and equipment lead times last many years. DLR controls scarce powered campuses and can serve six large hyperscalers across regions. Issuing equity at a rich public multiple to buy assets at above a 6.5% cap rate is accretive; 11.5% development yields are even better. The fee platform provides capital and management income, while a low payout retains cash. Per-share FFO compounds at 9–11%, validating the multiple.

Strongest bear case: the market is paying the 97th percentile of DLR’s history and approximately the same cash-flow multiple as Equinix for a business with weaker captivity. The capital-light story is narrative more than funding reality: DLR issued $2.5 billion through the ATM, bought back Blackstone’s stake, increased Teraco ownership, bought powered land, and acquired a manager. The company is building into record sector capital, and the apparent 11.5% yield is projected, not realized. When power bottlenecks ease, wholesale returns and the multiple can compress together.

Variant view: both popular sides miss the split. The operating bull is correct now; the imminent-collapse bear is not. But the stock is priced as though operating success automatically becomes per-share compounding. DLR’s own history says that inference is unsafe. The variant is to underwrite the denominator — diluted economic units and external funding — as rigorously as bookings and megawatts.

12. Fact vs. Interpretation

13. Open Questions

1. What is the transaction-level Core FFO/share accretion bridge for Blackstone after assumed debt, remaining capex, interest, and stabilization timing?
2. What fee-related earnings, margins, carried-interest economics, and retention compensation did Columbia bring for the upfront and contingent consideration?
3. How many Columbia shares were ultimately issued, and will earnouts be paid in cash or stock?
4. Will DLR recycle the Northern Virginia assets into a fund, or has it permanently reversed part of the capital-light strategy?
5. Do current projects realize the disclosed 11.5% yield after all cost and timing slippage, especially in EMEA?
6. Can annual diluted economic-unit growth fall below 3% after the transaction package closes?
7. Does physical cross-connect growth accelerate above 2.8%, proving the higher-captivity product is taking share?
8. Does 0–1 MW renewal pricing remain near 5%, and does LTM blended pricing remain above 6% after the Singapore comparison rolls off?
9. When will Kansas City secure a customer, and what total cost and stabilized return accompany the 600 MW/2 GW power option?
10. Will H2 recurring capex reach the $400–425 million full-year guide, or will AFFO remain flattered by maintenance timing?

14. What Must Be True — Falsification Scorecard

Bull case

1. Clean Core FFO/share compounds 8–10% beyond 2026. Current score: green; clean Q2 grew around 10% and FY midpoint implies 10.6% reported ex-promote growth. Falsified by two consecutive quarters below approximately 5% or a guide cut toward low-single digits.
2. Realized development returns remain double-digit. Current score: green/amber; expected yield is 11.5%, not yet realized. Falsified by new-start or realized yields below 9–10% after full cost.
3. Renewal pricing stays positive. Current score: green; LTM cash spread is 11%. Falsified by blended spreads approaching zero or the 0–1 MW layer turning negative.
4. The private-capital platform reduces common-equity dependence. Current score: red; H1 issuance accelerated. Confirmed only when diluted economic-unit growth stays below 3% while fee-related earnings rise.
5. The valuation can be earned through per-share growth. Current score: unproven; multiple remains near peak. Falsified if the multiple de-rates while clean FFO/share growth cannot offset it.

Bear case

1. Power scarcity is temporary and supply catches up. Current score: not yet; U.S. vacancy remains extremely tight. Falsified near term by continued sub-2% vacancy and sustained double-digit lease spreads; validated by local vacancy and pricing normalization.
2. Development returns compress toward 8–9%. Current score: not yet; expected yield rose to 11.5%.
3. Dilution prevents per-share growth. Current score: not yet; dilution is high, but clean per-share growth remains about 10%. Validated if unit growth remains above 5% and FFO/share stalls.
4. The multiple returns toward 16–20x. Current score: not yet; stock remains near its high.

The scorecard is split: operating tests favor the bull; capital discipline favors the bear; valuation offers no cushion.

15. Source Appendix — Principal Evidence

Primary sources were preferred; third-party aggregates were reconciled to filings where material.

1. Digital Realty, Q2-2026 Form 10-Q, filed 2026-07-31 — financial statements, acquisition accounting, share issuance, debt, JVs, Teraco denominator treatment, Item 5.
2. Digital Realty, Q2-2026 earnings release and supplement, filed 2026-07-23 and corrected 2026-07-31 — FFO/AFFO, bookings, renewals, guidance, pipeline, capex, debt, customer concentration.
3. Digital Realty, Q2-2026 presentation, 2026-07-23 — leasing, backlog, development, strategic priorities.
4. Digital Realty, Q2-2026 earnings call, 2026-07-23, via ROIC.ai; load-bearing figures cross-checked to the filing package.
5. Digital Realty, June transaction 8-K and release, 2026-06-22 — Kansas City, Teraco, Columbia terms.
6. Digital Realty, Blackstone transaction 8-K and release, 2026-06-29 — consideration, leases, cap rate, stabilization.
7. Digital Realty, Columbia Capital 424B7, filed 2026-08-19 — close, resale registration, earnout and lock-up details.
8. Digital Realty, 2026 DEF 14A, filed 2026-04-17 — ownership, compensation metrics, carried-interest plan.
9. Digital Realty, FY2025 Form 10-K, filed 2026-02-13 — history, customer concentration, fee income, FFO, debt and risk factors.
10. CBRE, Global Data Center Trends 2026, published 2026-06-17 — inventory, vacancy, absorption, pricing by market.
11. Equinix, Q2-2026 Form 10-Q, filed 2026-07-29 — peer financials, customer concentration and capital plan.
12. EDGAR Form 4 corpus, 2021-08-20 through 2026-08-20 — 288 filings and 355 transactions; two 2022 code-P buys, none since September 2022.
13. AZI valuation_index, dated 2026-08-19 — DLR P/B 97.275th percentile, P/S 97.395th, composite 81.669th; GAAP P/E ignored.
14. FactorsToday, dated 2026-08-19/20 — factor loadings, relative strength, drawdown and related-stock positioning.
15. Five-year adjusted price CSV through 2026-08-19 — event map, record high, and period returns.
16. ROIC.ai fundamentals — supplementary profile, balance-sheet and valuation cross-checks. Q2 missing/garbled fields were rejected in favor of filings.

APPENDIX A — Standard Diligence Questionnaire

General and Earnings Nature

What is the central investor debate? Whether current power scarcity and AI demand can produce durable per-share compounding, or whether a capital-intensive wholesale model again grows assets and revenue faster than each owner’s claim. The operating evidence is stronger than the capital-allocation evidence.

Are earnings at a cyclical high or low? Absolute Core FFO/share is early in a recovery from a flat 2021–2024 period. Renewal spreads, scarcity, and valuation are closer to cyclical highs. A recovering earnings base can coexist with a top-of-cycle multiple.

How stable is revenue? High. Most revenue is contractual rent and interconnection income. The principal risk is concentration and renewal economics rather than customer churn in ordinary conditions.

Is growth externally or internally driven? Both. External AI/cloud demand and grid scarcity drive leasing and pricing. Internal execution, the land/power bank, balance-sheet repair, and private-capital platform determine how much reaches each share.

Business Quality and Moat

What are the barriers to entry? In retail colocation: network density, metro location, cross-connect ecosystems, customer switching costs, power and operating reputation. In hyperscale: secured power, land, permits, equipment, capital and execution. The first set creates captivity; the second mainly delays supply.

Can customers switch or self-build? Retail customers face meaningful operational switching costs. Large hyperscalers can multi-source or self-build, although power scarcity and speed-to-market currently make leasing attractive.

Does brand matter? Reliability and global delivery matter more than brand. PlatformDIGITAL helps sell a consistent global platform; connectivity density is the harder asset.

How does DLR compare with EQIX? DLR has more direct hyperscale/AI-training exposure and a larger powered-land option. EQIX has the stronger network moat, more than twice the interconnections, lower customer concentration, and better demonstrated development economics.

Financial Condition

What is the correct earnings metric? Core FFO/share and AFFO/share, carefully adjusted for promotes, insurance, recurring capex, and dilution. GAAP EPS and ROIC are distorted by real-estate depreciation and gains.

How capex-hungry is the business? Extremely. H1 real-estate improvements exceeded operating cash flow before acquisitions and JV contributions. AFFO is distributable cash before growth capex, not conventional free cash flow.

Is AFFO conservative? Directionally useful but optimistic. It deducts defined recurring capex and external leasing costs, excludes growth capex, and adds back internal leasing compensation. Full-year maintenance guidance must be checked against actual H2 spending.

Is the balance sheet safe? Yes at present: 4.7x leverage, 5.2x fixed-charge coverage, 90% fixed and 92% unsecured debt. Higher new-debt costs and the size of the build are forward risks, not current distress.

Are there off-balance-sheet exposures? DLR owns equity-method stakes and manages development vehicles. Pro-rata JV debt was nearly $2.0 billion at June 30. Partner capital reduces consolidated funding but not the economic exposure of DLR’s retained stakes, fees, and commitments.

Capital Allocation and Management

How is cash used? The $4.88 dividend consumes roughly 64% of clean AFFO; retained cash, ATM equity, debt, asset recycling, and partner capital fund development and acquisitions.

Does DLR repurchase shares? No. It is a large net issuer and had $6.3 billion of ATM capacity remaining at June 30.

Was the Blackstone acquisition sensible? Potentially. Equity issued at a 4.29% FFO yield bought contracted assets at above a 6.5% stabilized cap rate. The absence of a complete accretion bridge prevents a definitive answer.

Are incentives aligned? Better than average because Core FFO/share, same-store NOI, and relative TSR matter. They are incomplete because no metric penalizes dilution or requires returns on incremental capital.

What does insider activity say? Little positive. There have been no open-market purchases since September 2022 and none under the current CEO. The prior report’s claim of zero purchases over five years was too broad and is corrected here.

Valuation and Market Data

What does the current price assume? Sustained high-single-digit or better per-share FFO growth, double-digit development returns, positive renewal pricing, successful fee-platform scaling, declining future dilution, and no material multiple compression.

Why ignore GAAP P/E? Real-estate depreciation suppresses earnings and disposition gains make them volatile. P/Core FFO, P/AFFO, EV/adjusted EBITDA, dividend yield, and own-history P/S are more informative.

Is the dividend attractive? It is covered, but a 2.54% yield is low for DLR history and the dividend has been frozen since 2022. The freeze is sensible because it retains capital.

Is DLR cheaper than EQIX? Not meaningfully on current clean cash-flow and EBITDA multiples. That is the strongest valuation objection because the franchise-quality gap still favors EQIX.

Risks and Recent Events

What can cause the stock to fall? Higher long rates, a guide cut, new-start yields below 10%, spread normalization, a hyperscaler pause, EMEA lease-up weakness, continued high dilution, or a broad AI-capex digestion that compresses the multiple.

What can make the analysis too cautious? If DLR realizes 11%+ yields, converts backlog on time, compounds clean FFO/share near 10%, and reduces unit growth below 3%, the company will have proven that its funding model can create durable value despite wholesale concentration.

What were the most important recent events? The June transaction package, the Blackstone buyout, the July Q2 results and guide raise, and the August Columbia close. Together they strengthened growth visibility while weakening the claim that the business is becoming asset-light.

APPENDIX B — Extended Source and Method Notes

The 60-month EDGAR sweep covered 491 filing-index rows, including 288 Forms 4, 90 Forms 8-K, 15 Forms 10-Q, five Forms 10-K, and five proxies. The local mirror passed its supplied integrity check with no zero-byte files. Retained duplicate filenames mean the manifest, not the raw directory count, is the completeness control.

The Q2 supplement was parsed at the table level because several aggregate financial feeds returned missing or malformed June fields. Filing figures control whenever they conflict with ROIC.ai or market-data snapshots. Current market price came from the adjusted price feed and was cross-checked to FactorsToday and AZI. The valuation percentile is an own-history rank, not a comparison with other companies.

Derived adjustments are transparent: the ex-promote/ex-insurance Q2 Core FFO sensitivity subtracts the $188 million net promote and $27 million insurance recovery from reported Core FFO, divided by the disclosed diluted shares/units. The AFFO sensitivity applies the same subtraction to the AFFO reconciliation. The clean H1 payout uses the $2.44 dividend divided by approximately $3.79 clean AFFO. Pro-forma share counts are sensitivities, not forecasts. The Teraco put shares were included in GAAP EPS/Nareit FFO but excluded from Core FFO/AFFO before settlement; settlement adds both shares and the related earnings claim.

Expected development yield is labeled as a management estimate. It is not a realized ROIC, and the greater-than-6.5% Northern Virginia acquisition cap rate is not substituted for it. The Equinix 27% figure is a realized return on a mature cohort and is used only as directional evidence of the product-mix quality gap.