Aligned Data Centers

Phase A — Understand the business

Lens 1 · Company Overview

What it is. Aligned Data Centers is a wholesale / hyperscale data-center developer-operator — it designs, builds, powers, cools and operates large-scale facilities and leases capacity (measured in megawatts of critical IT load), not square feet or racks, to hyperscalers, "neocloud" AI providers, and large enterprises on long-term (typically 10–15yr) contracts. It is the classic "landlord of the AI economy" model: capital-intensive real assets + power procurement + proprietary cooling, monetised as long-duration contracted cash flows. This is a REIT-like infrastructure business in economics (long leases, heavy capex, asset-backed debt), even though it is privately held rather than a listed REIT like Equinix or Digital Realty.

Scale (all ``, buyer-consortium disclosures + factboxes).

• ~5 GW of "operational and planned" / "contracted and available" capacity. Caveat: this blends operational + under-construction + planned — the operational base is materially smaller (Macquarie's own history: Aligned grew from 85 MW operational in 2018 to "over 5GW operational and planned" by 2025 ).
• ~50 data-center campuses across the Americas.
• Footprint: US core markets — Northern Virginia, Dallas, Phoenix, Salt Lake City, Chicago, Ohio, Maryland — plus Latin America (Brazil/São Paulo, Mexico/Querétaro, Chile/Santiago, Colombia) via the 2023 ODATA acquisition.
• HQ: Plano/Dallas, Texas. Founded 2013.

Customers. Publicly, Aligned names its customer base as "hyperscalers, neocloud, and enterprise innovators". Specific named tenants are sparse (this is a confidentiality-heavy business): the Factbox cites Nutanix and Datto as customers, and Aligned has a new Texas facility under development with NVIDIA-backed Lambda — a telling signal, since Lambda is a GPU-cloud (neocloud) buyer. Concentration is almost certainly high and hyperscaler-weighted (industry-wide, hyperscalers are ~65% of operating colo capacity as of Q1 2026, up from ~45% in early 2022 ) but the actual tenant split is not disclosed — n/a — private, not disclosed.

Contract structure. Long-term leases on critical IT load; industry norm for a comparable 2026 hyperscale deal (Applied Digital's Delta Forge) was ~$7.5B total contracted value over ~15 years for ~300 MW — i.e. ~$0.5B/yr per 300 MW, a useful sighting shot for what Aligned's per-MW annualised revenue looks like, though Aligned's own terms are undisclosed.

Bottom line: a top-tier US wholesale platform (routinely listed alongside Equinix, Digital Realty, QTS, CyrusOne, Switch, Vantage as a market leader ) whose differentiation is speed-to-scale + proprietary cooling + a green-financing track record, now valued as a core AI-infrastructure asset.

Lens 2 · Supply Chain

Map: inputs → Aligned → end customer. Named stakeholders where sourceable; generic where the private disclosure is thin.

Upstream (what Aligned buys / depends on):

• Power — the binding constraint. Aligned procures grid interconnection + PPAs; in constrained markets (Texas especially) the industry is shifting to behind-the-meter natural gas "shadow grid" generation (~101 GW announced BTM gas in the US, ~38 GW of it in Texas). Aligned's LatAm/Brazil assets run ~85–90% renewable via ODATA's structure. Power availability, not capital, is now the throttle on its growth.
• Cooling — proprietary, in-house: the patented Delta Cube (Delta³) fan-wall heat-rejection array (one moving part, supports up to ~50kW/cabinet) and the Cactus heat-rejection system. This is a genuine supply-chain internalisation — Aligned doesn't buy its cooling edge from CRAC-vendors, it builds it, and does not sell it to competitors.
• Land + grid-connected sites — controlling shovel-ready, interconnection-secured land is the scarce input; "developers who control grid-connected, shovel-ready land negotiate from a different position".
• Power/electrical + mechanical gear — generic industry chain (the census peers are the vendors): switchgear/UPS/busway (Vertiv, Eaton, Schneider, ABB), gensets (Cummins, Generac, Caterpillar), transformers (GE Vernova, Hubbell, Powell), thermal (Trane, Johnson Controls, Modine, Vertiv). Aligned is a buyer across this stack; none is single-source at the platform level.
• Construction / EPC — data-center GCs and electrical contractors (Quanta, Sterling Infrastructure, IES Holdings, Comfort Systems, MYR Group are the listed proxies).
• GPUs/servers — Aligned generally does not own the compute (unlike a neocloud); tenants bring their own NVIDIA GPUs. So Aligned is less directly exposed to GPU depreciation than CoreWeave-style operators — its risk is the tenant's willingness/ability to keep paying the lease, not the resale value of the silicon.

Downstream (who Aligned sells to):

• Hyperscalers (Microsoft/Amazon/Google/Meta/Oracle class — the anchor tenants), neocloud/GPU-cloud buyers (Lambda named; CoreWeave-class), large enterprises. The buyer syndicate itself (Microsoft, NVIDIA, xAI inside AIP) creates a strategic tenant-adjacency: the new owners are also among the largest buyers of exactly this capacity.

Chokepoints / single-source dependencies:

1. Power interconnection — the #1 chokepoint; 5–6yr grid queues in key markets. Flagged as a critical-path risk in the acquisition itself ("power interconnect timelines identified as critical path").
2. Cooling IP is a strength, not a chokepoint — internalised.
3. Tenant concentration — hyperscaler-weighted demand means a small number of counterparties drive the book (undisclosed, but structurally concentrated).

Names present → lens satisfied. The one number that matters: power is the binding input, and it is the explicit gating risk on the deal closing and on the ~5GW pipeline converting.

Lens 3 · Competitive Advantages (moats)

1. Speed-to-scale ("build-to-scale") + proprietary cooling. Aligned's pitch is delivering large blocks of high-density capacity fast, with the Delta Cube enabling ~50kW/cabinet and letting customers "add power capacity instead of more space". In a market where time-to-power is the scarce commodity, a developer that can energise dense capacity quickly has real pricing/allocation leverage. Moat type: process + IP + execution, moderately durable.
2. Capital access / balance-sheet firepower. Aligned has been a serial, record-setting capital raiser: first-ever US data-center sustainability-linked financing ($1B facility), first-ever green data-center securitization ($1.35B, Q3'21), a sustainability-linked loan scaled $375M → $1.75B, and a >$12B raise in Jan 2025 ($5B primary equity + $7B debt). Post-deal, it sits inside AIP's $30B-equity / up-to-$100B-with-debt vehicle — arguably the deepest capital pool in the sector. In a business where capital is the raw material, this is the strongest moat.
3. Green/ESG financing brand — genuine first-mover credentials that lowered its cost of capital and appealed to ESG-mandated hyperscaler procurement. Durable-ish but commoditising as peers copy.
4. Land + interconnection bank — sites with secured power in tier-1 markets (NoVa, Texas, Phoenix) are a scarce, appreciating asset.

Bargaining power. Over suppliers: moderate-high — a ~5GW buyer of gear/EPC has scale leverage. Over customers: structurally weaker than it looks — hyperscalers are giant, sophisticated, multi-source counterparties who play developers against each other; the tightness (~1.0–1.6% vacancy, record-low ) currently tilts power toward the landlord, but that is a cyclical advantage, not a structural one. The moat is real but not a wide, permanent one — this is infrastructure, and infrastructure competes on cost of capital, power access, and execution, all of which well-funded rivals (Vantage, QTS/Blackstone, CyrusOne/KKR-GIP, CloudHQ, Switch) also have.

Lens 4 · Segments

No segment financials are disclosed — Aligned is private and reports nothing. n/a — private, not disclosed on revenue/EBITDA by segment.

What can be segmented qualitatively:

• By geography: US (the majority — NoVa, Dallas, Phoenix, SLC, Chicago, Ohio, Maryland) + Latin America (Brazil, Mexico, Chile, Colombia via ODATA, 2023). At ODATA-close Aligned described >2.5 GW of critical capacity across 40+ DCs at full build-out for the combined Americas platform; by the 2025 sale that had grown to ~5 GW / ~50 campuses. Directional read: US is the core and the fast-grower (AI demand); LatAm is the greener, lower-multiple diversification leg.
• By product: effectively single-product (wholesale critical-IT-load leasing), with a shift up the density curve — from traditional colo toward AI-optimised, liquid-cooling-ready, 50kW+/cabinet builds. The trend is accelerating toward AI/high-density, which is the whole investment case.
• By customer type: shifting from enterprise/traditional toward hyperscaler + neocloud — mirroring the industry move to ~65% hyperscaler share.

Trend + cause: capacity roughly doubling every ~4 years industry-wide; Aligned's own trajectory (85 MW → ~5 GW in ~7 years, "65x growth" under Schaap ) is faster than market — a share-gainer, funded by relentless capital raising.

---

Phase B — Measure performance

Lens 5 · Funding & Valuation Trajectory (+private swap — replaces "Earnings Result")

This is the spine of the private story. Round/valuation history (all ``, unaudited):

Valuation read. The headline is a ~$40B EV — described by multiple outlets as the largest data-center transaction in history, exceeding Switch's 2022 ~$11B take-private. Discrepancy flagged: one LinkedIn post cited "$61.5B" — this is not corroborated by the official Macquarie/CNBC/Reuters figure of ~$40B EV; I anchor on ~$40B and treat $61.5B as an unverified single-source outlier (possibly conflating EV-plus-future-committed-capital, or simply wrong). Do not use $61.5B.

Burn / capital intensity signal. The Jan-2025 raise ($5B equity + $7B debt) nine months before a $40B sale tells you two things: (1) this is a capital-devouring build machine — it needed $12B of fresh fuel to keep the ~5GW pipeline moving; (2) Macquarie chose to sell at the top of the cycle rather than keep funding the buildout itself — a rational infra-PE exit after a ~7yr hold, not a distressed sale.

Lens 6 · Founder / Management Signal (+private swap — replaces "Earnings Calls")

No earnings calls (private). The equivalent signal is founder/executive posture + who is buying.

• CEO Andrew Schaap (since 2017): consistently frames Aligned as an infrastructure-supercycle play — "technologies such as AI, IoT, VR and blockchain will call for considerably more compute" — and post-deal says the AIP partnership will "accelerate our mission to deliver the infrastructure powering tomorrow's digital economy". Tone: expansionary, execution-focused, capital-markets-fluent. Schaap stays on as CEO through the ownership change — continuity signal.
• The loudest sentiment signal is the buyer list. The consortium is a who's-who of the AI stack: BlackRock GIP (lead), MGX (Abu Dhabi sovereign), inside AIP — Microsoft + NVIDIA (founders), with xAI, Cisco, GE Vernova, NextEra as partners, and Kuwait Investment Authority + Temasek as anchor LPs. When the two largest buyers of AI capacity (Microsoft, NVIDIA-ecosystem) and the deepest infra allocators on earth pay ~$40B for a landlord, that is the strongest possible institutional endorsement of the demand thesis — and simultaneously a conflict/vertical-integration flag (owners ≈ customers).

What they've stopped saying / watch for: the ESG/sustainability messaging that defined 2021–2023 (green securitization, 100%-renewable-by-2024 KPI) is quieter in the 2025 AI-era framing — the narrative has shifted from "greenest colo" to "fastest, densest AI infrastructure at scale." A subtle but real repositioning as power availability trumps carbon-purity in the AI land-grab.

Lens 7 · Cap Table & Secondary Marks (+private swap — replaces "Comps"), with a peer-valuation table

Cap table quality — top of the syndicate spectrum. Post-close ownership: BlackRock GIP (lead) + MGX + AIP (Microsoft, NVIDIA founders), seller Macquarie Asset Management. This isn't crossover-fund-markup territory (Fidelity/T.Rowe) — it's a strategic take-private by the sector's dominant capital+demand axis. The presence of Microsoft and NVIDIA on the ownership side is an unusually strong IPO-proximity/quality tell — except the exit here isn't an IPO, it's absorption into a permanent-capital infra vehicle.

Peer-valuation table. Aligned itself has no public multiple (private, financials undisclosed) — so its own row is a transaction-implied estimate, and peers are the listed comparables.

Sector benchmarks for triangulation: data-center M&A averages ~$12M/MW at ~20–25x EBITDA. Aligned at ~$40B on ~5GW is ~$8M/MW if you count all planned capacity — which flatters the multiple. On operational MW only (a smaller, undisclosed base), the implied $/operational-MW is well above the $12M/MW average, i.e. the buyers paid a full-cycle, growth-priced number for the platform + pipeline + team + power bank, not a trailing-cash-flow number. This is a bet on the pipeline converting, not on today's EBITDA.

Secondary marks: n/a — private, no active secondary market disclosed. The $40B take-private is the mark.

Lens 8 · Funding / Product Catalysts (+private swap — replaces "Stock-Price Catalysts")

The events that "moved the value" (there's no ticker, so these are the value-inflection points):

• 2018 Macquarie control — institutional capital unlocks the build machine (85 MW base).
• 2021 green securitization + SLL scaling — proved a novel, cheap, ESG-branded funding channel; lowered cost of capital.
• 2023 ODATA — stepped the platform to >2.5 GW and into LatAm; scale re-rating.
• Jan 2025 $12B raise — the fuel injection that made the ~5GW pipeline credible.
• Oct 15 2025 the $40B deal — the terminal value event; crystallised ~7yr of compounding into the largest DC deal ever.
• H1 2026 expected close — the remaining binary: CFIUS + regulatory approval is the one thing that can still move (or break) the value.

Pattern: value in this name is driven by (1) capital-markets events and (2) scale milestones, not by quarterly operations — exactly what you'd expect for a private infra platform. The market (here, the buyer syndicate) reacts to secured capacity + power + capital access.

---

Phase C — Judge people & books

Lens 9 · Management

• Andrew Schaap — CEO (since 2017), Board member. Track record: grew Aligned ~65x under his tenure (85 MW → ~5 GW), led the first international acquisition (ODATA) and the QScale (Canada) strategic investment, and executed "several historic capital raises". Pedigree: ex-Digital Realty SVP running global large-scale client builds, part of the exec team that grew DLR revenue to ~$2B; 25+ yrs across data centers, IT, PE and real estate. Archetype: professional operator-dealmaker (not a technical founder) — precisely the profile that fits a capital-intensive, M&A-and-financing-driven infra roll-up. Strong, well-matched.
• Meghan Baivier — CFO (since 2024). 23+ yrs finance/leadership; came from Easterly Government Properties (REIT) where she was President & COO. A REIT-native CFO installed right before the $12B raise and the $40B sale — a deliberate capital-markets hire to run exactly those processes. Signals the board was positioning for a large financing/exit well ahead of the October deal.
• Founder: Aligned was founded 2013; the founder handed the CEO role to Schaap in 2017 to "build for scale" — a clean founder→professional-manager transition, appropriate for the capital scale-up phase.
• Capital allocation: the entire history is capital allocation — raise cheaply (green/SLL innovation), deploy into land+power+builds, acquire (ODATA, QScale), and exit to the highest-capitalised buyer at cycle top. On the evidence, disciplined and value-creating (a ~7yr Macquarie hold ending in the largest sector deal ever is a strong IRR outcome, though the actual return is undisclosed). ROE/ROIC: n/a — private, not disclosed.
• Red flags (governance): the sharpest is owner-customer conflict of interest post-close — Microsoft, NVIDIA (via AIP) and MGX will own a landlord they and their ecosystems rent from, raising related-party/transfer-pricing and preferential-allocation questions that a public company would have to disclose but a private one need not. Also heavy leverage ($7B+ debt in one 2025 raise alone; asset-backed securitizations) — normal for infra, but a rate/refinancing exposure. No promotional-CEO or fraud signals surfaced.

Lens 10 · Forensic Red Flags

No audited financials exist (private, no SEC filings) — so a line-by-line forensic read is not possible; this is itself the headline caveat. ``. What a forensic analyst flags structurally for a business like this:

• Revenue recognition / lease accounting — long-term take-or-pay-style leases can front- or back-load recognised revenue; with no disclosure, unverifiable. Watch for how "contracted" ~5GW is defined (signed leases vs. LOIs vs. planned) — the ~5GW "operational and planned" phrasing deliberately blends secured and speculative capacity, which flatters the scale narrative.
• Capacity definition risk — the gap between operational and "operational + planned" MW is the single biggest place optimism hides. The ~$8M/MW implied multiple depends on counting planned capacity that requires power and capital not yet secured.
• Leverage / off-balance-sheet — extensive securitizations and SLLs; asset-level debt structures common in infra can obscure consolidated leverage. Refi risk if rates stay high or the AI-capex narrative cracks.
• SBC — n/a in the public-non-GAAP sense (private), but management/rollover equity in the take-private is a related-party consideration.
• Cash-flow vs. earnings — a build-phase infra developer burns cash (capex >> operating cash flow) by design; the Jan-2025 $12B raise confirms it. That is expected, not a red flag per se, but it means the business is valued on future stabilised cash flows, not current ones — the classic place valuations overshoot.

Regulatory findings (required sub-section).

• SEC (EDGAR EFTS — LR + AAER): None possible. Aligned has no CIK — it is private and not required to file with the SEC. total_sec_findings: 0. No Litigation Releases or AAERs can name it.
• Non-SEC enforcement (web search): No material FTC/DOJ/FDA/CFPB/consent-decree/fine findings surfaced for "Aligned Data Centers". The relevant regulatory exposure is prospective, not historical: the CFIUS review of the MGX/sovereign-capital acquisition (Abu Dhabi + Kuwait + Temasek buying US AI-critical infrastructure) is an active national-security screen on the deal itself. This is a deal-completion risk, not a finding of wrongdoing.
• Item 3 (Legal Proceedings): n/a — no 10-K exists (private).
• Verdict: No material regulatory or legal findings — verified via SEC EDGAR EFTS (no CIK, 0 findings), web search (no enforcement hits), as of 2026-07-06. The one live regulatory item is forward-looking: CFIUS approval of the acquisition (sovereign/foreign buyers of US AI infrastructure), unresolved pending H1-2026 close. All company financial claims are unaudited per public sources.

---

Phase D — Project & stress-test

Lens 11 · IPO-Readiness & Path-to-Tradeable (+private swap — replaces "Forward Projection")

This is the be-early payoff lens — and for Aligned the answer is unusual: the path-to-tradeable is NOT an IPO.

• Stage: pre-IPO in the census sense, but functionally under-acquisition — a signed take-private at ~$40B EV, expected to close H1 2026.
• IPO readiness (1–5 scale): I score it 2 — low, and getting lower as a public-IPO prospect, because the October 2025 deal removes it from the IPO path entirely — it is being absorbed into BlackRock/GIP's permanent-capital AI Infrastructure Partnership, not prepped for an S-1. Infra platforms held by GIP/BlackRock/MGX are long-hold private assets; they don't typically IPO on a near horizon. (Contrast: a name like CoreWeave did IPO; Aligned's owners have the opposite model.)
• Milestones that unlock a tradeable event:

1. CFIUS + regulatory clearance → deal close (H1 2026) — the immediate binary.
2. Thereafter, the realistic "tradeable" exposures are indirect: (a) BlackRock (BLK) and NVIDIA (NVDA) equity as listed proxies for the syndicate; (b) any future securitizations / infra bonds Aligned issues (a credit, not equity, access point); (c) a very-long-dated eventual re-listing or sale — not a 2026–2027 event.

• Catalyst (for the census/private-watch entry): Deal close H1 2026 pending CFIUS; thereafter a long-hold GIP/AIP asset — value already crystallised at ~$40B EV, no near-term public float expected.
• rNPV / EPS projection: not applicable — no EPS to forecast (private, no listed equity), and no forecast.ts Brier line is logged (per --watchlist rule + no committed base case on an untradeable asset). The "valuation" is the transaction: ~$40B EV, which the market has already set.

Write-back note: research/private-watch.json has no aligned entry — this dossier flags that gap. A future maintenance pass should add aligned with beat: datacenters, stage: under-acquisition, ipo_readiness: 2, catalyst: "$40B take-private by AIP/MGX/BlackRock GIP; CFIUS-pending H1-2026 close; long-hold private thereafter", dossier: this file so privates.ts ranks it correctly. (Per wave boundaries I do NOT edit that file here — this is the note for the human/next pass.)

Lens 12 · Bull vs Bear

Bull case. Aligned is arguably the single best private proxy for the AI-infrastructure supercycle. It owns ~5GW of the scarcest asset in tech — grid-connected, high-density, hyperscaler-ready capacity — in the tightest market on record (~1–1.6% vacancy ), with a proprietary cooling edge (50kW/cabinet Delta Cube), a serial cheap-capital machine, and now the deepest capital pool in the sector (AIP's $30B equity / $100B-with-debt) plus its two biggest customers (Microsoft, NVIDIA-ecosystem) as owners. Demand is secular: global DC capex heading >$600B in 2026 (+36% YoY), hyperscaler capex $635–690B, capacity doubling every ~4 years. A landlord with secured power and land in this environment prints long-duration, inflation-linked, contracted cash flows — and the $40B mark says the smartest infra money on earth agrees.

Bear case (permanent-impairment risks).

1. AI-capex is a cycle, and this is priced for the peak. 75% of hyperscaler capex is AI infrastructure; AI assets depreciate ~20%/yr, and 2025's implied hyperscaler depreciation ($400B) already exceeds their combined profit. If hyperscalers cut or pause (Amazon FCF projected negative in 2026; capex at 45–57% of revenue — utility-like ), Aligned's fastest-growing demand cohort throttles back, the ~5GW "planned" pipeline strands, and a $40B growth-priced valuation compresses hard. Multiple analysts flag mid-decade (2025–26) as the potential oversupply inflection where order-rate surpasses actual consumption.
2. Power is the throttle and the cost. 5–6yr interconnection queues; the pivot to behind-the-meter gas raises cost, carbon, and execution risk. Any planned-MW that can't get powered is worth far less than the multiple assumes.
3. The moat is cyclical, not structural. Landlord bargaining power over giant hyperscaler tenants is a tight-market phenomenon; in a downturn, well-capitalised rivals (Vantage, QTS, CyrusOne, CloudHQ, Switch — all similarly funded) compete pricing down. Infrastructure ultimately competes on cost of capital, and everyone in this cohort now has cheap capital.

Pre-mortem (18 months out, thesis broke): It's early 2028. A 2026–27 AI-capex air-pocket (a model-progress plateau + hyperscaler ROI scrutiny) hit just as record supply delivered. Vacancy backed up from ~1% toward high-single-digits; lease rates rolled over; several "planned" GW never got powered or funded. The $40B mark looks like a top-tick, refinancing the securitized debt got expensive, and the owner-customer conflict (Microsoft/NVIDIA renting from themselves) drew antitrust/related-party scrutiny. Aligned the asset is fine; Aligned the 2025 valuation was the mistake.

Are multiples too high? On operational cash flow, almost certainly yes vs. the ~$12M/MW, ~20–25x-EBITDA sector average — the buyers paid for the pipeline + team + power bank + strategic control, not trailing EBITDA. That's defensible for a strategic permanent-capital owner and dangerous for a financial one.

Contrarian view (what the market refuses to see): The consensus treats the $40B deal as a bullish signal for the whole sector. The contrarian read: it may be a top signal. Macquarie — a savvy infra-PE seller — chose to exit, not keep building, at cycle peak, offloading the ongoing capital burden onto sovereigns and strategics who must own capacity for reasons other than pure financial return (Microsoft/NVIDIA/MGX are buying strategic access + demand-security, not necessarily the best risk-adjusted return). When the most price-sensitive money sells and the least price-sensitive money buys, that's often the late innings.

Lens 13 · Devil's Advocate (short-seller)

(No public equity to short — so this is "what breaks the asset / would you underwrite the $40B?")

• Structural break: the business is a leveraged bet on hyperscaler capex staying vertical. Revenue is concentrated in a handful of mega-tenants (undisclosed but structurally so). If 1–2 anchor hyperscalers renegotiate, delay, or right-size AI buildouts (entirely plausible if AI unit-economics disappoint), a huge chunk of the "planned" pipeline evaporates and the growth-priced valuation has no floor.
• Most dangerous competitor bulls underestimate: not the other landlords — it's the hyperscalers self-building. Microsoft, Amazon, Google, Meta increasingly build their own campuses (and Microsoft is now literally a co-owner of Aligned — so it can build or buy). If self-build accelerates, third-party wholesale demand growth slows exactly when supply peaks. Vantage (freshly $3.6B-funded ) and QTS/Blackstone (>90% pipeline-to-secured-contract ) are the sharpest pure-play rivals.
• Capital-allocation / governance: the owner-customer conflict is the accounting/governance soft spot — related-party lease terms between AIP owners (Microsoft, NVIDIA) and the asset they own are undisclosed and unauditable, and could either flatter Aligned's book (sweetheart anchor leases to justify the price) or become an antitrust/transfer-pricing liability.
• Assumptions that must hold for $40B: (1) AI-capex compounds through 2027+; (2) the ~5GW "planned" largely converts to powered, leased MW; (3) interconnection/power gets solved at acceptable cost; (4) rates don't blow up the securitized debt stack; (5) no hyperscaler demand air-pocket. Break any two and the mark is impaired.
• If growth disappoints 20–30%: on a growth-priced ~$40B (paying for planned capacity), a 20–30% haircut to the demand/conversion assumption plausibly takes fair value well below the transaction price — the buyers, as permanent-capital strategics, would hold through it, but a financial underwriter should not pay $40B.
• Single scenario that permanently impairs: a sustained AI-capex retrenchment (model-progress plateau → hyperscaler ROI discipline → capex cuts) coinciding with record 2026 supply → oversupply, falling lease rates, stranded planned GW, and expensive refis. Plausibility: moderate and rising — the depreciation math (~$400B/yr > profits) and the "mid-decade inflection" warnings are real, not fringe.

Lens 14 · Management Questions (ordered by information value)

1. Of the ~5 GW "operational and planned," how many MW are operational and cash-flowing today, how many are contracted/leased-but-not-built, and how many are planned with power not yet secured? (The single number that determines whether $40B is cheap or a top-tick.)
2. What is the tenant concentration — revenue share of the top 1, 3, and 5 customers — and the weighted-average remaining lease term? Are the leases take-or-pay?
3. Post-close, what are the governance guardrails on related-party leasing to AIP owners (Microsoft, NVIDIA) and their ecosystems — arm's-length pricing, independent review, disclosure?
4. How much of the pipeline has secured, energised power vs. sitting in interconnection queues, and what is your behind-the-meter generation strategy/cost (gas, and on what carbon timeline)?
5. What is consolidated net leverage (including all securitizations/SLLs), the weighted-average cost and maturity wall of the debt, and refinancing exposure if rates stay elevated?
6. What are stabilised unlevered yields-on-cost for recent US builds, and how have they trended as land/power/construction costs rose?
7. What is the churn/renewal experience — have any hyperscaler tenants delayed, downsized, or renegotiated commitments in the last 12 months?
8. How do you underwrite AI-demand durability — what's your base/bear case for hyperscaler capex through 2028, and what triggers a build-pace cut?
9. What is the CFIUS mitigation structure, and what happens to the pipeline/financing if approval is delayed or conditioned?
10. What is the capex plan and external funding requirement for the next 24 months to deliver the pipeline, and where does it come from (AIP equity, securitization, project debt)?
11. How defensible is the Delta Cube cooling advantage as liquid cooling standardises industry-wide, and are you exposed on water in Phoenix/Texas?
12. What is the LatAm (ODATA/QScale) contribution to revenue/EBITDA and its growth vs. the US core — is it a drag or a diversifier?
13. What are the KPIs the AIP owners will judge you on — MW delivered, yield-on-cost, tenant quality, or strategic capacity-security for the owners themselves?
14. What is the realistic long-term liquidity path for this asset (re-IPO, sale, perpetual hold), and on what horizon?
15. Where would you most want a competitor to overbuild, and which market are you most worried is heading to oversupply in 2026–27?

---