Digital Realty Trust (DLR) — Refresh Delta (since 2026-06-21)

PRE-PRINT NOTE. DLR reports Q2 2026 results 2026-07-23, after market close — tomorrow. Nothing below is a Q2 result. This dossier is a pre-print setup refresh: what changed in the primary record since the last dossier, what the Street is pricing in, and what to watch when the print lands. Grounding: 10-Q (Q1 2026, filed 2026-05-01, already on shelf) re-read in full for leasing/development-pipeline detail the prior dossier under-sourced; three new 8-Ks (2026-06-22, 2026-06-29, 2026-07-01) pulled fresh from SEC EDGAR via curl -A "MenFem Research [email protected]"; regulatory findings re-run (still 0 SEC findings). No 10-Q/10-K/8-K since 2026-06-21 changes the FY2025/Q1-2026 financial spine — the moves below are M&A/capital-markets activity and Street positioning, not restated financials.

---

Refresh delta (since 2026-06-21) — the 5 load-bearing facts

1. DLR bought out Blackstone's 64% interest in two Northern Virginia hyperscale JVs (Dulles 9 + Brickyard) for ~$3.58B, closed 2026-06-30. 288 MW across 3 data centers, 100% pre-leased to three distinct investment-grade hyperscale tenants, 15-year leases at a blended Aa3/AA- rating and 3.6% annual escalators, >6.5% initial stabilized cap rate. ``
2. DLR disclosed a quantitative, dated power-delivery commitment for a new site — Astra Enterprise Park, Kansas City: 600 MW of utility power contracted by early 2028, rising to 2 GW at full delivery, via an Energy Service Agreement with the local utility. `` This is the single highest-value item in this refresh for the KB's interconnect-queue gap — see §2 below for the important caveat.
3. The Teraco put flagged as a red flag in the 2026-06-21 dossier is now executing. DLR agreed 2026-06-22 to issue 3,425,031 shares (~$644–650M) to buy ~16% of Teraco from third-party put-holders, raising its stake from ~61% to 77%. ``
4. DLR's own Q1 2026 10-Q — already on shelf but under-mined in the last dossier — shows the development pipeline nearly doubled in one quarter and the pre-lease/yield metrics both softened slightly: 1,169 MW under construction as of 2026-03-31 (+52% QoQ vs 2025-12-31's ~769 MW), 61% pre-leased (down from 64% at 2025-12-31), ~11.4% blended stabilized yield (down from 11.9% for FY2025). ``
5. The hyperscaler capex backdrop DLR's thesis rides on moved materially higher. The prior dossier cited ~$600B of 2026 hyperscaler capex; the current aggregate across Amazon/Google/Microsoft/Meta guides to ~$725B (+77% YoY), with Microsoft separately disclosing an $80B backlog of Azure orders it cannot fulfill because of power constraints — an external, non-DLR data point that corroborates DLR's own "power is the binding constraint" framing at the hyperscaler level. ``

What held. The core structural facts from 2026-06-21 are unchanged and not re-litigated here: single-reporting-segment structure, the >1MW/0-1MW revenue split (~60/35), the NoVA/power land-bank thesis, the private-capital flywheel mechanic. See Lenses 1–4 and 7 in deep-dive-2026-06-21.md.

Position-seed move. None taken. Per this refresh's Socratic-gate instruction, no directional call, bull/bear verdict, or position-seed language is written this pass — this is a raw-data delta only. No MarketCall exists yet for DLR (the prior dossier's Position Seed was WATCHING/BULLISH but was never promoted via /thesis).

---

1. Leasing — bookings, backlog, renewal spreads (priority 1)

All figures below are Q1 2026 (quarter ended 2026-03-31) — the latest actuals on file; nothing newer exists pre-print. The 10-Q table is the primary source; the press release / earnings-call figures cross-validate (or in one case, conflict with) it.

New leasing, DLR's share, by cohort ``:

Plus interconnection $18.6M → **total DLR-share new leasing $422.8M** (the "$423M" figure the prior dossier cited without full attribution). **At 100% share: $706.9M** — the second-highest quarterly leasing total in company history, ~70% above the next-highest quarter ``.

By region (DLR share, ex-interconnect) ``: Americas $320.9M (0-1MW $40.4M / >1MW $280.1M / other $0.4M) · EMEA $37.4M ($29.3M/$8.0M/$0.1M) · APAC $45.8M ($9.2M/$36.4M/$0.2M). Americas = ~79% of new bookings.

Largest single lease: a 200 MW AI-inference-oriented lease with a AA-rated hyperscaler in Charlotte — "the largest single lease in Digital Realty history," first hyperscale deployment for that metro ``. Counterparty not named.

Renewals, DLR's share ``:

Company-reported blended figures (a different cut than the table above): cash +5.0%, GAAP +6.3% on $193M of total renewals . The CFO separately stated **>1MW renewals achieved a 7.4% cash spread** — this conflicts with the +10.6% implied by the 10-Q's per-kW table above. Both are DLR-sourced; I am not reconciling the gap (likely a $-weighted vs kW-weighted or lease-mix difference) — flagging per provenance discipline rather than picking one. Worth a direct question tomorrow.

Backlog: $1.8B of annualized GAAP base rent at 100% share, $1.0B at DLR's share — signed-but-not-commenced, giving visibility into 2027 and 2028 . Weighted-average lag from signing to revenue commencement: **19 months** — a new data point, not in the prior dossier. This backlog figure has not moved since 2026-06-21 — it is Q1 2026 quarter-end data; whether it grew is exactly what tomorrow's print answers.

Additional color: 0-1MW + interconnection bookings totaled $98M, +40%+ YoY, a record; 21% of signings were AI-oriented requirements `` — the cleanest single stat for "how much of colo growth is genuinely AI-driven" in the file.

---

2. Power — contracted vs. available MW, grid-interconnect (priority 2)

The KB gap flag stands: DLR discloses committed power delivery dates for specific sites, but never a queue wait-time metric. I searched the full Q1 2026 10-Q, both June/July 8-Ks, the June 22 press release, and the June 22 investor deck for "queue," "interconnect," "grid," and found no wait-time disclosure anywhere. What DLR does disclose — and this is genuinely new and quantitative — is a contracted delivery timeline for a specific greenfield site:

Astra Enterprise Park, Kansas City — 1,440 acres acquired 2026-04-30 for $377.6M cash + 517,475 OP units (8-K language) / ~$475M all-in cash+units (press-release framing — both provenance-labeled, not reconciled) . DLR "entered into an Energy Service Agreement with the local utility for 600 megawatts of utility power to be provided by early 2028, and two gigawatts expected at full capacity" . The investor deck specifies the 600MW/early-2028 figure is **Phase 1, on the north parcel — 280 of the 1,440 acres** .

Read for the KB: this is a site-specific, utility-agreement-backed delivery date, not a queue wait-time (i.e., it doesn't tell you how long DLR waited in the interconnection queue to get this commitment, only what was committed and when it lands). It is nonetheless the most concrete power/timeline number in the file this cycle and directly responsive to the KB's stated gap — it should be logged as a positive datapoint on the "does the industry disclose delivery timelines at all" question, while the "how long is the queue" question remains genuinely unanswered by DLR's own disclosure. Recommend the KB record both halves of that distinction.

Other power color from the Q1 call ``: ~90% of utility expense is reimbursed by customers; the remaining ~10% is "largely hedged or contractually adjustable" — i.e. power-cost inflation is substantially insulated at the P&L level, consistent with the prior dossier's lease-structure finding. CEO Andy Power acknowledged "shortages of skilled labor, particularly electricians" as a build constraint (distinct from grid/interconnect capacity itself), while noting market rent growth is still outpacing build-cost inflation.

Carried unchanged from 2026-06-21 (not re-verified this pass, no new filing touches it): land bank >3,500 MW developable, >1,000 MW in Northern Virginia ``.

---

3. Development pipeline & capex — under construction vs. announced (priority 3)

Under construction, DLR's own 10-Q language: "As of March 31, 2026, we had 1,169 megawatts of projects underway... representing an approximately 52% increase of capacity under development as compared to December 31, 2025. As of March 31, 2026, 61% of the 1,169 megawatts... was pre-leased... we estimate that the stabilized yield on our total 1,169 megawatts of capacity under construction... was approximately 11.4%." — press/call materials round this to "1.2 GW."

This is a real delta from how the prior dossier characterized the pipeline. The 2026-06-21 dossier cited "769 MW underway... 64% pre-leased" and an "11.9% stabilized dev yield" — those numbers are the FY2025 (2025-12-31) figures from the 10-K, sourced there as `` rather than pulled from the Q1 10-Q that was already sitting on the shelf. The corrected picture, now research-layer-sourced: pipeline nearly doubled in one quarter (769→1,169 MW), pre-lease rate softened 3pts (64%→61%), and blended yield compressed ~50bps (11.9%→11.4%). None of this is alarming on its own (a bigger, slightly-less-pre-leased pipeline at a still-double-digit yield is what an accelerating build looks like), but it's a materially different starting point for tomorrow's number than the prior dossier implied.

Gross pipeline value: $16.5B under construction, up >60% from year-end 2025, ~80% in the Americas `` — a dollar-denominated figure not present in the prior dossier at all.

Future development capacity (unchanged bucket, now correctly research-layer-sourced): >5 GW total; 58% from sites with >100MW buildable, 30% from 25–100MW sites, 12% from <25MW sites ``.

New land acquired in 2026 (adds to the future-capacity bank; none of this predates 2026-06-21 except where noted):

• Kansas City / Astra Enterprise Park — 1,440 acres, 600MW/2GW power (detailed in §2 above) ``
• Atlanta — 873 acres, $95M, "expected to support over one gigawatt" `` — Q1 2026 activity, backfilled here as it wasn't in the prior dossier
• Portland — 30 acres, $50M, "expected to support 160 megawatts" `` — same

Blackstone JV buyout (closed 2026-06-30) — read carefully, this is a consolidation, not new pipeline. DLR already held a blended ~36% interest in the Dulles 9 and Brickyard JVs; it bought Blackstone's remaining 64% for ~$3.58B ($1,231M cash + ~$2,346.1M in non-voting common stock, ; press coverage rounds this to "$3.5B" ). The underlying assets: 288 MW across 3 hyperscale data centers in Northern Virginia — 2 in Manassas (96MW each, formerly 80% Blackstone) + 1 on the Dulles/Sterling campus (96MW, formerly 50% Blackstone) — 100% leased to three distinct investment-grade hyperscale customers, 15-year leases, blended Aa3/AA- credit, 3.6% annual escalators. Gross value $7.8B; implied initial stabilized cap rate >6.5%. Two centers stabilize H1 2027, the third H1 2028. Guided leverage-neutral and Core FFO accretive in 2027 and 2028. `` This increases DLR's wholly-owned exposure to already-under-development NoVA capacity — it does not add incremental MW to the >5GW future bank.

---

4. Customer concentration (priority 4)

Carried, unchanged (FY2025 10-K basis): top-20 = 50.9% of annualized rent; top-3 ≈ 26%; #1 ~11.7% ($547M); #2 Oracle 9.0% ($424M) ``.

New (Q1 2026 10-Q, different basis — flagged explicitly): "Our largest customer's total revenue approximates 11% of our total revenue base. No other individual customer makes up more than 10% of our total revenue." `` Note the basis mismatch — the 10-K figure is % of annualized rent; this 10-Q figure is % of total revenue. Directionally consistent (~11–11.7%) but not a strict like-for-like — I am not collapsing them into a single trend line.

New exposure since 2026-06-21: the Blackstone JV buyout adds three additional, wholly-consolidated hyperscale tenants (blended Aa3/AA-, 100% leased on the 288MW) — none named in the 8-K . Whether these are existing top-20 names (increasing concentration) or net-new logos (diversifying) is not disclosed and is a live open question for tomorrow's call. Separately, the **200MW Charlotte lease is a single unnamed AA-rated hyperscaler** — another concentration-relevant data point with no counterparty name attached .

---

5. Other material corporate actions since 2026-06-21 (capital markets / M&A)

• ATM issuance: 2026-04-29 through 2026-06-29, DLR sold 6,158,839 shares under its at-the-market program for ~$1.2B net proceeds, used to temporarily pay down revolver borrowings and fund development/working capital ``. This is in addition to the FY2025 $1.1B ATM cited in the prior dossier's red-flags lens — this single ~2-month window raised more than half of all FY2025 ATM activity, continuing (arguably accelerating) the dilution trend flagged there.
• Teraco put executing (see delta #3 above) — the prior dossier's contingent "Teraco put overhang" is now a signed, quantified transaction: 3,425,031 shares (~$644–650M) for ~16%, ownership 61%→77%, expected close 2H26 ``.
• Columbia Capital acquisition: ~$485–510M (cash + stock + up to $275–290M earnout through YE2027) to acquire 100% of Columbia Capital, a digital-infrastructure private-capital firm with >$9B in fund commitments across 16 vehicles ``. Columbia is a long-time Teraco co-investor and Vela Infrastructure partner — this deepens existing relationships rather than adding a new counterparty, and directly extends the "private-capital flywheel" mechanic (Lens 9 in the prior dossier).
• Blackstone's exit via secondary: on 2026-07-01, Blackstone sold all 12,310,249 shares of its converted non-voting stock in an underwritten public offering at $185.00/share; DLR received no proceeds ``. Standard mechanics for stock-as-acquisition-consideration, not necessarily a signal — noting it as a data point, not interpreting it.

---

Macro backdrop moved (feeds Lens 8 — stock-price catalysts, carried)

• Hyperscaler 2026 capex guide is materially higher than at the last dossier. Prior dossier: ~$600B (+36%). Current aggregate across the big four: Amazon $200B (from $125B), Google $175–185B (from $91B), Microsoft $110–120B (from $90B), Meta $125–145B (raised from $115–135B) → ~$725B collective, +77% YoY from ~$410B in 2025; 75% ($450B) explicitly AI-related ``.
• Microsoft disclosed an $80B backlog of Azure orders it cannot fulfill due to power constraints `` — an external, non-DLR data point directly corroborating "power, not demand, is the binding constraint," at the scale of DLR's largest customer.
• Stock price has drifted down into the print. 2026-06-21 (prior dossier): ~$190. 2026-07-22 (today): intraday $177.26–$180.80, prior close $180.44, after-hours ~$177.92 ``. For reference, the June capital-markets activity was priced well above today's level: the 6/22 stock-consideration deals at a $197.54 weighted average, the 7/1 Blackstone secondary at $185.00. Net: DLR has de-rated ~5–8% since the last dossier, into a print with a Street-guided sequential FFO dip.
• New sell-side coverage: BTIG initiated Buy, $215 PT, 2026-07-10 . Consensus PT ~$217–222 across 32–33 analysts (21 Strong Buy / 2 Moderate Buy / 10 Hold per one aggregator) .

---

What to watch tomorrow (2026-07-23, after close)

• Consensus Core FFO/share: $1.98 (Zacks, +5.9% YoY from $1.87) — **note this is sequentially below Q1's actual $2.04.** That is not obviously a deceleration signal: CFO Matthew Mercier told the Q1 call to expect **"a step down in the second quarter, starting to rebound in the third,"** attributed to opex timing and development-investment timing. A print at/near $1.98 would be in line with guidance, not a miss vs. trend — worth reading tomorrow's number against that explicit prior guardrail, not against Q1's absolute FFO.
• Consensus revenue ~$1.66B (+11.4–11.5% YoY); rental revenue ~$1.12B (+12.1%) ``.
• Beat history: topped consensus every quarter over the trailing 4 quarters, average beat +5.11% ``.
• Did backlog ($1.8B/100% · $1.0B/DLR-share) grow, and did the 19-month signing-to-commencement lag hold or stretch as the pipeline scales?
• Did the pipeline's 61% pre-lease / 11.4% yield (now on a much larger 1,169MW+ base) hold, improve, or continue softening — the QoQ move already flagged in §3 is the trend to extend or reverse.
• Any reconciliation of the 5.0%/10.6% vs. 6.3%/7.4% renewal-spread conflict flagged in §1 — a direct question worth asking on the call.
• Any interconnect/queue-wait-time disclosure — historically absent; a first would be a genuinely new KB data point.
• Full-year 2026 Core FFO guidance ($8.00–$8.10) — reaffirmed, raised, or trimmed for the newly closed Blackstone/Teraco/Columbia deals and the added ATM dilution.
• Management color on integrating the Blackstone NoVA assets, and whether the 3 unnamed hyperscale tenants on that 288MW are existing top-20 names or new logos.
• Any commentary connecting to the hyperscaler capex reset upward (~$725B) or Microsoft's $80B power-constrained Azure backlog — does DLR frame this as tailwind-confirming or does management stay guarded.

---

Carried forward — not re-run this refresh

Lens 1 (Company Overview), Lens 2 core supply-chain map (beyond the Kansas City power update in §2), Lens 3 (moats), Lens 4 (segment mix — unchanged single-segment structure), Lens 7 (Comps), Lens 14 (Management Questions) — unchanged since 2026-06-21, see deep-dive-2026-06-21.md.

Lens 11 (Forward Projection), Lens 12 (Bull vs Bear), Lens 13 (Devil's Advocate), and the Position Seed block are deliberately not run this refresh. This pass is scoped to data + mechanics under an explicit Socratic gate — no directional call, bull/bear verdict, or position-seed language. The 2026-06-21 dossier's Position Seed (BULLISH/WATCHING, MEDIUM conviction) stands as the last recorded model view; it has not been re-affirmed or revised here.

---

Provenance key: = 10-Q (Q1 2026, filed 2026-05-01) or 8-Ks (2026-06-22, 2026-06-29, 2026-07-01) pulled directly from SEC EDGAR this pass. = live search/fetch, dated. `` = own derivation, shown. One unresolved conflict flagged in §1 (renewal-spread cohort math) rather than silently reconciled. No forecast.ts created; no ingest-deep-dive.ts sync run (dispatched/unattended pass — publish stays with Connor). Regulatory findings re-verified 2026-07-22: still 0 SEC Litigation Releases / AAERs.