Vantage Data Centers

Phase A — Understand the business

Company Overview

What it is, in plain terms. Vantage builds and operates wholesale hyperscale data centers — it acquires large land parcels where power is available, develops campus-scale shells (typically 100+ MW per campus, now scaling to gigawatt mega-campuses), and leases whole buildings or large capacity blocks to a small number of very-large-credit tenants on long (10–15 year) contracts. It is a landlord-developer of AI/cloud infrastructure, not a cloud or a chip company. The revenue model is contracted, recurring rent on triple-net-style leases; the value creation is development spread (build at a yield-on-cost above the exit cap rate) plus capital recycling (securitize stabilised assets, redeploy into new builds).

Scale. ~35 campuses (operational + under development) across 5 continents, 14 countries, 21 markets. Existing + planned global capacity is >2.6 GW, with a development pipeline management describes as 3+ GW and "over 100% expansion in 36 months".

Customers / suppliers / competitors.

• Customers: ~80% of capacity is contracted to hyperscalers (large cloud operators), the balance to colocation/enterprise. Named recent counterparties: OpenAI + Oracle (Stargate — Texas "Frontier" and Wisconsin "Lighthouse"/Port Washington), and hyperscaler tenants across its German fleet (Frankfurt/Offenbach, Berlin — "fully leased to hyperscale customers"). Google and Microsoft appear in "land and expand" adjacency around Vantage sites.
• Suppliers: power (the binding input — see Lens 2), long-lead electrical gear (transformers, switchgear, generators), and increasingly on-site prime power partners — VoltaGrid (>1 GW natural-gas microgrid) and Liberty Energy (1 GW, incl. 400 MW reserved 2027 generation).
• Competitors: the wholesale/hyperscale set — Digital Realty (DLR), Equinix (EQIX), QTS (Blackstone), CyrusOne (KKR/GIP), Aligned, Compass, STACK, CoreWeave, Crusoe, Switch, NTT.

Contract structure / concentration. Long-dated wholesale leases with high-credit tenants — the feature that lets Vantage issue investment-grade asset-backed securitizations (see Lens 5/7). But the marginal growth is highly concentrated: the two flagship 2025-announced campuses (Frontier + Lighthouse, >$40B combined investment) are both OpenAI/Oracle Stargate capacity. That is the central tension of the whole business today (Lens 12/13).

Supply Chain

Map: land + power → Vantage (develop/operate) → hyperscaler tenant. Named stakeholders along the chain:

Upstream — power (the real product). In 2026 the scarce input is not concrete, it is megawatts on a schedule. ERCOT reported ~226 GW of large-load interconnection requests as of Dec 2025, up from ~63 GW a year earlier, ~three-quarters from data centers. Grid interconnection is now a multi-year queue, so Vantage has verticalised into on-site prime generation to bridge (or replace) the grid:

• VoltaGrid — partnership (Feb 2025) to deploy >1 GW of prime natural-gas microgrid power across Vantage's North America portfolio; ~70 MW/unit, deployable in months as a bridge or long-term source. (Same VoltaGrid that Oracle tapped for a 2.3-GW modular gas fleet — a shared, and therefore contended, supplier.)
• Liberty Energy — partnership for 1 GW of power solutions, with 400 MW of 2027 generation capacity reserved.

Upstream — long-lead equipment. Transformers, switchgear, generators, and now liquid-cooling systems (Frontier is designed for 250 kW+ racks with liquid cooling). Chokepoint: high-voltage transformers and gas turbines/gensets are industry-wide constrained; Vantage's VoltaGrid/Liberty reservations are a hedge against exactly this.

Midstream — Vantage. Land assembly (1,200 acres at Frontier; 672 acres at Port Washington), construction management, commissioning.

Downstream — tenants. Hyperscalers (Oracle/OpenAI at the flagships; unnamed hyperscalers on the German fleet). Beyond them sits the actual end-demand: AI training/inference workloads. Vantage does not touch the GPUs (Nvidia) — its tenant (Oracle) or the tenant's tenant (OpenAI) does.

Chokepoints / single-source dependencies: (1) Power delivery date — the whole model breaks if MW slip; the gas-generation partners are the single most important supply-chain hedge. (2) VoltaGrid contention — the same firm powering Vantage is powering Oracle directly (2.3 GW), so Vantage's gas bridge competes with its own customer's gas fleet for the vendor's throughput. (3) Tenant credit on the growth vintage (OpenAI) — see Lens 13.

Competitive Advantages (moats)

What the moat actually is. For a wholesale developer the moat is not brand or product differentiation (a hyperscaler's shell is a commodity to first order). It is (a) a landbank with secured/reservable power, (b) speed-to-energised-capacity, (c) balance-sheet firepower, and (d) sponsor relationships. Vantage scores on all four:

• Power-secured landbank + speed. The VoltaGrid/Liberty gas reservations let Vantage promise energised MW on a hyperscaler's timeline while grid-only competitors sit in the ERCOT queue. In 2026 time-to-power is the moat, and Vantage has explicitly bought it.
• Balance-sheet firepower. DigitalBridge + Silver Lake put $9.2B of equity (upsized from $6.4B, "significantly oversubscribed") into Vantage in 2024, sized to enable ~$30B of additional development. Plus €2.2B EMEA debt in 12 months and multiple ABS issuances (below). Capital is the moat when the constraint is capital + power.
• Securitization franchise. Vantage pioneered data-center ABS in new markets — first-ever EMEA data-center ABS (£600M sterling), then the first euro-based Continental-Europe data-center ABS (€640M, +€80M VFN). A repeatable, investment-grade financing machine (Class A-2 rated A-, Class B BBB- by S&P/Scope) lowers cost of capital vs. peers who fund on-balance-sheet only.

Bargaining power. Over suppliers: moderate — Vantage is a large buyer of power/equipment but competes with hyperscalers (who self-build) and neoclouds for the same gear and gas. Over customers: weak at the top end — its hyperscaler tenants are among the largest, most sophisticated buyers on earth, and increasingly self-build (AWS/Microsoft/Google build owned capacity, compressing third-party demand). Vantage's leverage is time: when a hyperscaler needs 1 GW now and can't wait for the grid, Vantage's ready-to-energise campus commands terms. That leverage is cyclical, not structural.

Durability verdict. Real but not wide. The scale/capital/power-reservation advantage is defensible for the current cycle; it is not a switching-cost or network-effect moat. The bear case (Lens 13) is precisely that the "moat" is a leveraged bet on a demand curve.

Segments

our figures does not exist (private, no disclosure) — so there is no segment table. From public sources, the de facto segmentation is:

• By geography: three platforms — North America (the growth engine: Texas Frontier, Wisconsin Lighthouse, Ohio, Quebec/Canada), EMEA (Germany fleet — Frankfurt/Offenbach + Berlin; the ABS-financed, "fully leased" core), and APAC (developing). The EMEA platform is separately capitalised (AustralianSuper €1.5B; GIC + MEAG €1.4B; ABS debt).
• By tenant type: ~80% hyperscale / ~20% colocation-enterprise.
• By vintage: stabilised, leased, financeable (Germany — the ABS collateral) vs. speculative/under-construction growth (Stargate campuses). This split is the most important one for risk: the securitizable book is diversified-hyperscaler; the growth book is OpenAI-concentrated.

Trend + cause. The mix is shifting hard toward North America AI mega-campuses — management guides ~25% capacity growth in 12 months (Ohio, Quebec completions) and >100% in 36 months via the 3 GW pipeline, "assuming continued hyperscaler commitments". Cause: the AI capexCapital expenditureMoney spent on long-lived things — buildings, machines, servers — rather than on running costs. supercycle and Stargate. The conditional ("assuming continued commitments") is the whole ballgame.

---

Phase B — Measure performance

Funding & Valuation Trajectory (+private swap for "Earnings Result")

No P&L is disclosed. The scoreable "print" for a PE-owned infra platform is its capital-raising cadence and cost of capital — and here Vantage's 2024–25 run is genuinely elite:

Reading the "beat." The $9.2B equity raise being upsized and oversubscribed at the top of a rate cycle is the single strongest performance signal in the file — sophisticated infra capital (DigitalBridge, Silver Lake, AustralianSuper, GIC, MEAG) underwrote Vantage's development pipeline with conviction. The A-/BBB- ABS ratings independently validate the lease book's credit quality. This is the bull case in one row.

Balance-sheet flags. Total funding stock cited at ~$23.3B — treat as approximate/unaudited. The structure is deliberately leveraged (development + ABS); a >$40B flagship capex plan against a private balance sheet means the model is exposed to (a) construction-cost/interest-rate shocks and (b) any tenant that fails to take delivery. No public covenant data.

Founder/Executive Interviews & Sentiment (+private swap for "Earnings Calls")

No earnings calls (private). Substituting management's public posture:

• Consistent narrative: "unprecedented AI demand," speed-to-power, mega-campus scale. The Frontier unveiling ($25B, 1.4 GW, liquid cooling, 250 kW+ racks) and the Wisconsin announcement are the set-pieces.
• What shifted 2024→2026: the story moved from diversified global wholesale growth to AI mega-campus + on-site power. The VoltaGrid (Feb 2025) and Liberty partnerships mark the pivot to power-as-a-service verticalisation — an admission that grid interconnection is now the binding constraint and that Vantage must own the bridge.
• Tone: relentlessly expansionary; no public hedging on demand. For a sponsor-owned platform in a capex land-grab that is expected — but it means there is no management signal that would warn of a demand air-pocket. The warning, if it comes, will come from the tenant (OpenAI/Oracle), not from Vantage.

Cap Table & Secondary Marks + Comps (+private swap)

Cap table / syndicate quality — a strong tell. The owners are tier-1 infrastructure/PE, not growth VCs: DigitalBridge (control) + Silver Lake (founder/co-lead), with crossover-grade institutional LPs — AustralianSuper, GIC (Singapore sovereign), MEAG (Munich Re). For a private, a GIC + sovereign/pension entry is the analog of a Fidelity/T-Rowe crossover markup: it signals institution-grade, IPO-or-strategic-exit-ready backing. No public secondary marks (PE-held, not on secondary markets like the AI unicorns).

Comps. No public multiple exists for Vantage itself (private). The relevant public reference set — data-center landlords — trades rich, which frames what a Vantage exit could fetch:

Provenance discipline: I will not fabricate a Vantage multiple. The honest read: public DC landlords trade at ~26–30× EV/EBITDA and 18–28× AFFO; the two closest structural comps (QTS, CyrusOne) are private, PE-owned — the exact playbook Vantage sits inside — so a strategic/continuation exit is the base-case liquidity path, not an IPO into a public multiple. Vantage implied EV: n/a (would require private EBITDA I don't have).

Funding / Product / Catalyst Events (+private swap for "Stock-Price Catalysts")

No stock price. The events that move Vantage's enterprise value (and would move a public proxy) over the last ~2 years:

• Jan 2024 — $6.4B equity announced → Jun 2024 completed/upsized to $9.2B: the capital that de-risked the pipeline. Value-up.
• Feb 2025 — VoltaGrid >1 GW power partnership: solved the binding constraint. Value-up.
• Aug 2025 — Frontier ($25B, 1.4 GW) unveiled; Oct 2025 — Wisconsin Lighthouse ($15B, ~1 GW) → >$40B Stargate backlog. Value-up, but concentration-up.
• 2025 — €640M euro ABS + €2.2B EMEA debt: cost-of-capital validation. Value-up.
• Late 2025–2026 — community/regulatory friction (Wisconsin noise + PSC energy-disclosure suit; Texas noise/water complaints): the emerging downside catalyst class.

Pattern: Vantage's value is driven by (1) capital access and (2) power reservations — both of which it has executed superbly — and increasingly gated by (3) tenant demand durability and (4) local permitting/social license. The market (were it public) would react most to any crack in (3): an OpenAI/Oracle Stargate slip or renegotiation.

---

Phase C — Judge people & books

Management

• CEO — Sureel Choksi (President & CEO since 2013; ~13-year tenure). Prior: operating executive at Silver Lake Partners (the founding sponsor), CEO of Elevation Data Centers, and a long run of senior roles at Level 3 Communications — CFO, CMO, President of Wholesale Markets, EVP Services, treasurer. Archetype: a wholesale-infrastructure operator + capital-markets executive, not a technologist or founder-visionary. That is exactly the right archetype for this business — the job is land, power, capital, and long-dated leases, and Choksi's Level 3 wholesale-markets + CFO pedigree is a direct fit.
• Track record — quantified: transformed Vantage from a single Santa Clara campus into a 35-campus, 5-continent, >2.6 GW platform over his tenure; presided over the $9.2B raise and the pioneering ABS program. This is a genuinely strong operational build.
• Skin in the game / ownership: control sits with DigitalBridge + Silver Lake; management holds equity/incentive units typical of a PE platform (specifics not disclosed — private). Insider ownership: n/a — private, not disclosed. Choksi's tenure and Silver Lake lineage suggest strong alignment with the sponsors.
• Capital allocation: the defining competence here. Vantage recycles capital textbook-style — build → stabilise → securitize (A-/BBB-) → redeploy — and raised equity at the right time (2024, ahead of the 2025 AI capex peak). No public evidence of value-destructive M&A or buybacks (not applicable to the structure).
• Red flags (governance): the usual PE-platform opacity — no public financials, related-party dynamics between Vantage / DigitalBridge (its manager and part-owner) / DigitalBridge's other vehicles are inherent and undisclosed. Not a scandal signal; a transparency limitation the analyst must price.

Verdict: high-quality, well-matched operator with a sponsor bench (DigitalBridge/Silver Lake) that is among the best in digital infrastructure. Management is not the risk here.

Forensic Red Flags

No audited financials, no P&L, no cash-flow statement are public — so the classic forensic tests (cash-flow-vs-earnings divergence, receivables/inventory outrunning revenue, SBC flattering non-GAAP) cannot be run. State that plainly: income-statement / balance-sheet / cash-flow forensics: n/a — private, unaudited, not disclosed. The analytically honest substitutes:

• Leverage/structure risk (the real forensic concern): the business is deliberately debt-heavy (development + ABS). The ABS structures ring-fence specific stabilised assets (e.g., the four German DCs) behind SPVs (Jersey Borrower SPV; Germany Borrower Lux). That is standard and rated — but it means the best collateral is encumbered, and the growth (Stargate) is funded against the residual + new equity/debt. An outside creditor to the growth book is structurally junior to the ABS holders on the stabilised book.
• Off-balance-sheet / JV opacity: EMEA is separately capitalised with AustralianSuper/GIC/MEAG; the consolidated economic picture is not public. Related-party layer (DigitalBridge as manager + owner) is undisclosed.
• Counterparty-quality bifurcation: the ABS collateral is diversified-hyperscaler (why it earns A-); the growth vintage leans on OpenAI/Oracle. A forensic analyst would flag that the rated credit quality of the existing book does not extend to the concentration risk of the new book — do not let the A- rating launder the Stargate exposure.

Regulatory findings (required sub-section).

• SEC (EDGAR EFTS — LR + AAER): regulatory/regulatory-findings.md (fetched 2026-07-07) records 0 SEC findings — Vantage has no CIK, is private, and is not required to file. No EDGAR enforcement search is possible against the operating company. (Note: the securitization SPVs file Form ABS-15G under CIK 0001727983 — routine ABS disclosure, not enforcement.)
• Non-SEC enforcement (web search): No material FTC/DOJ/FDA/CFPB enforcement action, consent decree, fine, or penalty against Vantage Data Centers surfaced. What did surface is local/civil friction, not federal enforcement: in Wisconsin, an advocacy group (reported as "MEA") sued the Public Service Commission (PSC) to compel energy-demand disclosure for the Meta (Beaver Dam) and Vantage (Port Washington) projects; Port Washington residents have protested 24-hour construction noise in council meetings. In Texas, community complaints about noise, dust, water use and potential contaminated-water discharge around Vantage construction. These are permitting / social-license / rate-case risks, material to timeline and cost, not solvency.
• 10-K Item 3 (Legal Proceedings): n/a — no 10-KForm 10-KA company’s audited annual report to the US regulator. The most complete thing it publishes. exists (private).
• Bottom line: No material regulatory or federal legal findings — verified via SEC EDGAR EFTS (LR, AAER: 0 hits), web search (no federal enforcement), and the absence of any 10-K to disclose litigation, as of 2026-07-07. The live risk is local opposition + energy-disclosure/rate litigation, which is escalating industry-wide and specifically names Vantage's Port Washington campus. Label all "unaudited per public sources."

---

Phase D — Project & stress-test

IPO-Readiness & Path-to-Tradeable (+private swap for "Forward Projection")

No EPS line is meaningful (private, no P&L) — so no our model forecast is logged (per --watchlist rules, and there is no scoreable EPS/binary anyway). The +private lens asks: what creates a tradeable event, and when?

Structural read. Vantage is a PE/infra-platform, so the liquidity paths, in order of base-case likelihood:

1. Sponsor exit via strategic sale or continuation vehicle (most likely). The direct comps — QTS (Blackstone, 2021) and CyrusOne (KKR/GIP, 2022) — were taken private, and large DC platforms are trading between mega-funds, not IPO-ing. DigitalBridge/Silver Lake could sell to a sovereign/pension/infra consortium or roll into a continuation fund. Estimated window: opportunistic, sponsor-driven — no fixed date.
2. IPO (possible but not the base case). Requires public markets to pay up for a leveraged, tenant-concentrated developer; the AI-datacenter-bubble debate makes the window regime-dependent. A public listing would most plausibly follow, not precede, Stargate lease stabilisation.
3. Partial monetisation via ABS/continued private debt (already ongoing) — recycles capital without a control event.

Readiness assessment: on the SKILL's 1–5 scale this maps to ~3 (late-stage / sponsor-active) — institution-grade owners, investment-grade financing, and scale sufficient for any exit, gated by (a) Stargate lease stabilisation and (b) the macro/AI-capex regime. This is not a near-term S-1. The single most important thing to do off this dossier: add Vantage to research/private-watch.json under datacenters with stage: late, ipo_readiness: 3, lead_investors: "DigitalBridge, Silver Lake, GIC, AustralianSuper, MEAG", catalyst: "Stargate (Frontier TX + Lighthouse WI) lease stabilisation → sponsor exit / IPO optionality", dossier: <this file> — closing the overlay gap flagged at the top. (Do not edit here — this is a --watchlist run; flag it for the operator.)

What I will not fabricate: Revenue: ~$700M/yr per one third-party source, but a second source says ~$287M — these conflict by >2×, both are low-confidence scrapes, so treat as directional only, not a basis for a valuation.EBITDA: n/a.Valuation: one scrape shows ~$2.24B — implausibly low for a >2.6 GW, $23B-funded platform (likely a stale/partial figure), so I discard it rather than repeat it. The disciplined output is: the enterprise is large and well-capitalised, but no reliable public EV/EBITDA/revenue figure exists — n/a, not sourced.

Bull vs Bear

Bull case. Vantage is the arms dealer of the AI build-out with the two things that matter most in 2026 — capital and power — already secured. $9.2B of oversubscribed sponsor equity, an investment-grade ABS machine (A-/BBB-), >2.6 GW in the ground/planned, and >1 GW of reserved on-site gas generation (VoltaGrid + Liberty) that lets it energise capacity while grid-only rivals wait years in the ERCOT queue. It sits at the center of Stargate — the single largest AI-infra program announced — with >$40B of flagship capacity contracted to OpenAI/Oracle. Ownership is elite (DigitalBridge, Silver Lake, GIC, AustralianSuper, MEAG); management (Choksi, 13 years) is perfectly matched. If AI demand compounds as bulls expect, Vantage is a prime, exit-ready beneficiary and its sponsors monetise into a rich DC-landlord multiple (public comps at ~26–30× EV/EBITDA).

Bear case (permanent-impairment risks).

1. Tenant concentration on an unprofitable counterparty. The growth vintage (Frontier + Lighthouse, >$40B) is underwritten by OpenAI/Oracle Stargate. OpenAI is cash-burning and its capacity commitments are ambitions, not stabilised triple-net leases. If OpenAI's funding or demand disappoints, Vantage is left with tens of billions of purpose-built, liquid-cooled, high-density capacity whose next-best tenant may not exist at that spec/location. This is the CoreWeave-62%-Microsoft problem in developer form.
2. AI-capex air-pocket / overbuild. 2025–26 is cited as a possible inflection where supply catches inflated order rates; three of four hyperscalers lost market value on capex worries. A leveraged developer is maximally exposed to a demand pause — build cost is sunk, leases on the growth book are not yet locked to stabilised credit.
3. Power/permitting cost + social-license drag. The gas-generation hedge invites emissions/permitting scrutiny; local opposition (Wisconsin PSC suit, Texas water/noise complaints) is escalating and can delay/raise the cost of the exact campuses driving growth.

Pre-mortem (18 months out, thesis broke — what happened?): OpenAI's Stargate ramp slipped or was renegotiated as its funding tightened; one flagship campus sat with energised-but-unleased capacity; rising rates + an AI-capex pause repriced DC-development risk; DigitalBridge's exit window closed; and the ABS-encumbered good assets couldn't rescue the un-stabilised growth book. Vantage didn't go bankrupt (infra assets have residual value) — but the equity return compressed hard and the "exit-ready" thesis slipped years.

Are multiples too high? Unanswerable for Vantage directly (private, no sourced multiple). But the reference set is priced for continued growth (EQIX ~74× P/E, ~30× EV/EBITDA) — i.e., a public exit assumes the AI build-out keeps compounding. That is the assumption to stress.

Contrarian view (what the market refuses to see): the consensus treats Vantage's A-rated ABS and blue-chip syndicate as if they de-risk the whole enterprise. They do not — they de-risk the stabilised German book while the value is being created on the Stargate book, which carries exactly the concentration/counterparty risk the ratings exclude. The credit quality and the growth risk live in different parts of the same company. Bulls conflate them.

Devil's Advocate (short-seller)

Dismantling the bull case:

• Where revenue is concentrated + what breaks it: the marginal, headline-grabbing growth (>$40B) rides on one program (Stargate) and effectively one demand source (OpenAI) intermediated by Oracle. Shift that concentration — OpenAI slows, or Oracle renegotiates its own take — and the growth thesis is impaired regardless of the German fleet's quality.
• Why the moat is weaker than bulls think: "speed-to-power via gas" is replicable — Oracle already contracted VoltaGrid directly for 2.3 GW, and every serious developer is racing to on-site generation. The landbank+capital+power edge is a cycle advantage, not a structural one; when capital is abundant (as now), it commoditises fastest.
• Most dangerous competitor bulls underestimate: the hyperscaler self-build (AWS/Microsoft/Google building owned capacity), which structurally shrinks the third-party wholesale TAM — and the tenant itself (Oracle) moving up the stack into its own power/campus development, disintermediating the landlord over time.
• Worst capital-allocation / governance concerns: not fraud — opacity and relatedness. DigitalBridge is both manager and owner; there are no public financials to check whether development yields-on-cost still clear the exit cap rate as construction and power costs inflate. The short's question isn't "is it a scam" — it's "at what yield-on-cost is Vantage now building, and does it still beat its cost of capital after the gas-generation and liquid-cooling premium?" No one outside can answer that.
• Assumptions that must hold for today's (implied) valuation: (1) AI demand keeps compounding through 2028+; (2) OpenAI/Oracle take delivery on Stargate at contracted terms; (3) power gets delivered on schedule at modeled cost; (4) rates/exit multiples for DC platforms stay rich enough for a lucrative sponsor exit. All four.
• If growth disappoints 20–30%: a leveraged developer with sunk build cost and un-stabilised growth leases sees equity value fall more than 20–30% (operating + financial leverage), and the exit window slips — the classic infra-development downside convexity.
• Single scenario that permanently impairs: OpenAI funding/demand shock that cascades through Stargate while rates are elevated — stranding purpose-built, location-specific, liquid-cooled capacity that can't be re-tenanted at spec. Plausibility: non-trivial and rising — OpenAI's economics are the most-debated question in the sector, and Vantage has concentrated its 2025-vintage growth precisely there.

Management Questions (ordered by information value)

1. What is your weighted-average lease term and tenant credit mix on the growth (Stargate) book specifically — and how does it differ from the ABS-financed stabilised book? (Highest-value: separates the A-rated collateral from the concentration risk.)
2. For Frontier and Lighthouse, are the OpenAI/Oracle commitments binding triple-net leases with the same credit standing as your German hyperscaler leases, or development/framework agreements — and what are the take-or-pay / termination terms?
3. What yield-on-cost are you underwriting new AI mega-campuses at today, and how much has it compressed as land, power, gas-generation, and liquid-cooling costs have risen — versus your blended cost of capital?
4. If OpenAI's Stargate demand slipped 25–30%, what is the re-tenanting plan for purpose-built, high-density, location-specific capacity, and what haircut to yield would that imply?
5. How much of the >$40B flagship capex is committed/funded today vs. contingent on future capital raises, and what happens to the plan if the equity/ABS markets tighten?
6. On power: how much of your 3 GW pipeline has firm, contracted energisation (grid + VoltaGrid + Liberty) vs. reserved-but-not-firm — and what is your exposure to gas-turbine/genset delivery slippage?
7. What is the all-in economic (not just accounting) exposure to gas generation — fuel-price, emissions-compliance, and permitting risk — as gas becomes central to your delivery model?
8. Given DigitalBridge is both manager and owner, how are related-party terms (fees, co-investment, asset transfers) governed to protect minority/co-investors?
9. What is the realistic liquidity path and horizon for DigitalBridge/Silver Lake — strategic sale, continuation vehicle, or IPO — and what milestones (Stargate stabilisation?) unlock it?
10. How exposed is the growth book to hyperscaler self-build compressing third-party wholesale demand over the next 3–5 years?
11. What are your consolidated leverage and interest-coverage metrics across all platforms (NA + EMEA + APAC), including the ABS SPVs and JV debt?
12. How are you managing the Port Washington / ERCOT permitting and rate-case / energy-disclosure litigation, and what is the timeline/cost risk to the flagship campuses?
13. What is your actual, audited group revenue and EBITDA — public third-party estimates range from ~$287M to ~$700M and are clearly unreliable?
14. How fungible is a half-built Stargate campus to a different tenant or workload (training→inference, one hyperscaler→another) if the anchor tenant changes?
15. What is your development-to-stabilisation timeline on the AI campuses, and at what point does the growth book become ABS-financeable (i.e., de-risked to the standard of the German fleet)?

---