AirTrunk

Phase A — Understand the business

Lens 1 · Company Overview

AirTrunk is APAC's largest hyperscale data-centre developer-operator — the region's answer to a QTS or a CyrusOne, but purpose-built from day one for the handful of buyers who take capacity by the hundred-megawatt. Founded 2015 by Robin Khuda (ex-CFO of NextDC and Pipe Networks) in Sydney, the thesis was singular: the hyperscalers (AWS, Azure, Google, Meta, ByteDance) were about to need enormous, cheap, fast, standardised capacity across Asia-Pacific, and no local operator was building at that scale or cost point. AirTrunk built to that spec and rode it.

What it actually is: a build-to-suit landlord of power and space. It develops large campuses, leases them under long-dated (10–15 year), inflation-linked capacity contracts where the customer pays for reserved power and space (a take-or-pay-like structure — you pay for the megawatts you booked whether or not you fill them). That contract structure is the whole investment case: it converts volatile AI-capex demand into bond-like, indexed cashflows with blue-chip counterparties.

Scale (the headline growth story):

• ~3.3 GW of operating + planned capacity across 22 campuses in 6 regions — Australia, Japan, Singapore, Malaysia, Hong Kong, India. (Note the ramp: ~2 GW / 14 campuses was the figure at the 2024 sale; the jump to 3.3 GW / 22 reflects the aggressive post-Blackstone expansion, especially India.)
• >1.4 GW operating-or-committed, of which >90% is leased to blue-chip tenants as of FY2025.
• Described by the company / press as ~2× the size of its nearest APAC competitor.

Customers: the global hyperscalers — Amazon, Microsoft, Google are named as tenants; ByteDance/TikTok is a confirmed, structurally important anchor (JV partner on a second Singapore site; anchor tenant across Malaysia). Singapore's EDB awarded AirTrunk (alongside Equinix, Microsoft, GDS) one of the scarce pilot-programme allocations.

Suppliers: power utilities & grid operators (the binding input), construction/EPC contractors, electrical gear (switchgear, transformers, gensets), cooling (increasingly liquid/direct-to-chip for AI density), and renewable developers for PPAs (OX2, ib vogt).

Owners: Blackstone (~88%, across four strategies — infrastructure, real estate, tactical opportunities, private equity) + CPP Investments (~12%), acquired Dec 2024 for an implied EV >A$24B (~US$16.1B) from Macquarie Asset Management + PSP Investments. Founder Robin Khuda remains CEO and retained a stake.

Lens 2 · Supply Chain

Map (upstream input → AirTrunk → end customer), every named stakeholder that is sourceable:

Upstream — the binding constraint is POWER, not silicon.

• Grid & utilities (the true chokepoint): national/state grid operators in each market. In Australia (NEM), Japan (TEPCO region for Tokyo campuses), Malaysia (TNB — Tenaga Nasional), Singapore (SP Group, capacity-rationed), India (state discoms + Maharashtra government for the Raigad site). Grid connection queues and power availability — not capital — are what gate the buildout. Johor is already rejecting up to ~30% of DC applications on power/water grounds.
• Renewable power (PPAs to hit Net Zero 2030): OX2 (25 MW solar, Riverina NSW — a tripartite Google/AirTrunk/OX2 24/7-CFE deal); ib vogt (30 MW solar, Malaysia's first DC virtual PPA).
• Construction / EPC: local hyperscale-experienced contractors (specific GCs not consistently disclosed — n/a — private, not disclosed).
• Critical electrical & mechanical gear: transformers, switchgear, UPS, generators (a global supply-constrained category in 2025–26), and cooling — pivoting toward liquid/direct-to-chip as AI rack density climbs past what air can handle.

Midstream — AirTrunk: land assembly + entitlement → power procurement → design/build of standardised hyperscale campuses → lease-up under long-dated contracts → operate.

Downstream — end customers (the hyperscale buyers): AWS, Microsoft Azure, Google Cloud, ByteDance/TikTok (anchor), plus other large internet/AI platforms. These are simultaneously AirTrunk's customers and its most dangerous potential competitors (they self-build — see Lens 3/13).

Chokepoints / single-source dependencies:

1. Power availability & grid connection — the master constraint; a campus with a signed lease and no energised grid connection is a stranded asset.
2. Water — now a live regulatory chokepoint in Johor (dedicated DC water tariff RM5.33/m³; a 100 MW site ≈ a 10,000-person town's daily water).
3. Customer concentration — a small number of hyperscalers = enormous buyer power (see Lens 3).

Lens 3 · Competitive Advantages (moats)

The moat is real but of a specific, bounded kind — it is an execution-and-scale moat, not a pricing-power moat.

• Scale & first-mover density in APAC (strong). ~2× the nearest competitor; the only operator with a genuinely pan-APAC hyperscale footprint (6 markets). For a hyperscaler wanting to land 100–300 MW across Tokyo + Sydney + Johor + Mumbai on one counterparty's paper, AirTrunk is close to the only call. That multi-market, single-relationship optionality is the differentiator.
• Cost & speed-to-build (strong, original thesis). Founded explicitly on a lower-cost, faster, standardised hyperscale template — the reason it won share off incumbents.
• Land + power bank (strong, and the scarce asset). Entitled land with secured/queued power in constrained metros (Tokyo, Johor, Singapore, Mumbai) is the genuinely scarce, hard-to-replicate resource. This is what a buyer is really paying for.
• Switching costs (moderate). Once a hyperscaler installs fleets of racks and fibre into a campus, physical migration is painful — but the leases are finite (10–15 yr) and the customer holds renewal leverage.
• Capital access under Blackstone (strong, and arguably the biggest post-2024 upgrade). Blackstone is positioning to be the largest AI-infra investor in the world (QTS grew >900% under it); that balance sheet + AirTrunk's >A$18B sustainable-finance platform (incl. the A$16B ex-Japan facility + US$1.24B Tokyo green loan) means AirTrunk can fund a buildout few rivals can match.

Bargaining power — this is the moat's soft spot. AirTrunk needs the hyperscalers more than any single hyperscaler needs AirTrunk. The customers are ~US$1–3T companies that can (and do) self-build, and in 2026 they are explicitly favouring flexibility over long-term lock-in in power-constrained markets. AirTrunk's power over suppliers (utilities, gear) is likewise weak — it is a price-taker on the scarcest input. Net: the moat protects share and returns on already-built capacity well; it does not confer pricing power over customers.

Lens 4 · Segments

No segment P&L is disclosed (private; segments.csv empty) — so this is a geographic capacity breakout, ``, not a revenue/EBITDA-by-segment table (n/a — private, not disclosed for segment economics):

Trend & cause: the mix is shifting hard toward India + Malaysia (SEA) — the two markets absorbing hyperscale spillover from constrained Singapore and expensive Australia. India went from zero to the single largest planned commitment (US$30B/5 GW) in one quarter (Q2 2026) via the Lumina deal — accelerating, driven by (a) India's sovereign AI/cloud demand and (b) Blackstone's capital enabling a land-grab. Japan remains the mature cash-generative core. The direction of travel is unambiguously more capacity, more geographies, funded by cheap sustainable debt — the classic Blackstone "buy a platform, then pour capital into its development pipeline" playbook.

---

Phase B — Measure performance

(+private overlay: Lens 5 → Funding & valuation trajectory; Lens 7 → Cap table & secondary marks; plus a Traction & unit-economics note. All ``, unaudited.)

Lens 5 · Funding & Valuation Trajectory (private overlay)

The equity story is a textbook infra-scaleup step-up ladder, seed-of-scale (2017) → 5× exit (2024):

Valuation step-up: ~A$3B (2020) → ~A$24B (2024) = ~8× in ~4 years, tracking the AI-capex supercycle re-rating of digital infrastructure.

Traction & unit economics (the operating reality behind the equity marks):

• >90% of >1.4 GW operating-or-committed capacity is leased to blue-chip tenants — very high utilisation of built/booked stock.
• Contracted capacity ~8× since 2020; footprint 5 → 11+ campuses over the same window.
• Profitability — CONFLICTING FIGURES, surfaced not reconciled:
• ~US$1B run-rate EBITDA at FY2024 acquisition → ~20–23× EV/EBITDA on the A$24B EV. This is the multiple the market anchored on.
• vs. AirTrunk's own forward guidance at deal announcement: ~A$340–350M revenue / ~A$210–220M EBITDA for "the current financial year".
• My read ``: these are almost certainly different bases — the ~US$1B is a stabilised, full-contracted-portfolio run-rate (or a forward-looking figure Blackstone underwrote), while the A$210–220M is a single in-progress fiscal year on the then-smaller operating base. A ~5× gap can't be a rounding error; a portfolio that has grown contracted capacity ~8× since 2020 plausibly spans that range across "current-year booked" vs "fully-stabilised run-rate." I am not confident which the REIT prospectus will use as its distributable income base — this is the single most important number to verify when the S-1/prospectus drops, and it directly sets the REIT yield.
• (A stray getlatka figure of "$208M ARR / $624M valuation" is stale/low-confidence scrape junk — discarded, flagged so it isn't mistaken for signal.)

Burn / balance sheet: development-heavy, so free cash flow is deeply negative by design (capex >> operating cashflow during the buildout) — funded by the >A$18B green-debt platform + sponsor equity. This is normal for a growth infra platform and is why Blackstone (patient, levered infra capital) is the right owner. n/a — private for net-debt specifics.

Lens 6 · Founder / Management Communication (private overlay — no earnings calls)

No earnings calls exist (private). Signal comes from founder interviews, press, and the cadence of announcements — which is the tell here. In H1 2026 AirTrunk has announced, in rapid succession: India entry + Lumina acquisition + US$30B/5 GW India plan (Jun), a 2nd Osaka campus (OSK2), two more Johor campuses (US$3B), and the Tokyo US$1.24B green loan (Mar), all while prepping the Singapore REIT IPO (Apr–Sep).

Tone / focus: relentlessly expansionary, AI-demand-framed, capital-markets-fluent. Khuda communicates as a builder in land-grab mode with a patient balance sheet behind him. The strategic through-line is unmistakable: build the platform as large as possible on cheap sustainable debt, then monetise a slice via the public REIT while retaining the private growth engine. That is a Blackstone value-creation script, executed publicly. No tonal cracks visible; the risk is that the announcement drumbeat outruns the grid's ability to energise it (see Lens 13).

Lens 7 · Cap Table & Secondary Marks (private overlay — comps by structure, not P/E)

Cap table quality (the be-early tell): top-tier.

• Blackstone (~88%) + CPP Investments (~12%). This is the IPO-proximity signal — a mega-PE sponsor + a crossover-grade pension both underwriting, with an explicit public-markets monetisation path (the S-REIT). Earlier syndicate (Goldman, TPG, Macquarie, PSP) was already institutional; the current owners are the highest-conviction combination for a datacentre platform on the planet.
• Founder Robin Khuda: retained stake; net worth ~US$2.1B (Forbes real-time, 2026); realised ~A$1B (~US$672M) of value at the 2024 sale. Skin in the game intact, still CEO.

Comparable transactions & marks (the right "comp table" for a private — deal multiples, not equity multiples):

Secondary / crossover signal: the cornerstone-investor solicitation for the S-REIT (approaching select investors ahead of an ~Aug 2026 launch) is the classic pre-IPO crossover step — an IPO-proximity tell in its own right.

Lens 8 · Funding & Product Catalysts (private overlay — events that re-marked the equity)

The "what moves the mark" pattern for a private is the funding/expansion event cadence:

• Dec 2024 — Blackstone/CPPIB close (A$24B). The defining re-rating; triggered a re-pricing of listed ASX DC names too (NEXTDC).
• Mar 2026 — US$1.24B Tokyo green loan (largest DC financing in Japan).
• Apr 2026 — India entry via Lumina CloudInfra (600 MW pipeline) + bank mandate for Singapore REIT.
• Jun 2026 — US$30B / 5 GW India plan; ~US$21B Raigad/Mumbai campus LOI.
• Aug–Sep 2026 (pending) — Singapore S-REIT IPO (~S$1.5B raise) — the next hard catalyst and the first time public investors can touch any of this.

What the pattern reveals: the mark is driven by (1) capital-markets access (each mega-financing lowers cost of capital and de-risks the buildout) and (2) new-geography TAM unlocks (India was a step-change). Not by quarterly operating beats — this is an infra-development story where securing power + land + cheap debt in new metros is the value-creation event.

---

Phase C — Judge people & books

Lens 9 · Management

Robin Khuda — Founder & CEO.

1. Track record (excellent, quantified): built AirTrunk from a 2015 startup to a >A$24B platform in ~9 years — one of the great infra-scaleup outcomes in Australian history. Prior operator credibility as CFO of NextDC and Pipe Networks — i.e. he'd already built and financed data-centre and telecom infra before founding his own. This is a domain-native operator-financier, not a generalist.
2. Tenure & skin in the game: founder, still CEO through three ownership regimes (Goldman/TPG → Macquarie/PSP → Blackstone/CPPIB). Retained a stake worth ~A$1B at the 2024 sale; ~US$2.1B net worth. Alignment is strong and he stayed through the sale — a positive signal the sponsors wanted continuity and he wanted the next leg.
3. Capital-allocation history (strong): repeatedly raised the right capital at the right time (Goldman/TPG growth equity 2017 → Macquarie infra 2020 → Blackstone mega-infra 2024) and matched it to a disciplined build-to-suit model with pre-leasing. The green-debt platform (>A$18B) is best-in-class infra financing. The open question is whether the US$30B India plan is disciplined capital allocation or land-grab over-reach (see Lens 13).
4. Red flags: none material surfaced. The ByteDance dependence is a strategic/geopolitical exposure rather than a governance red flag. As a private under Blackstone, governance/comp are not publicly disclosed (n/a — private).
5. Archetype: domain-expert founder-financier — the ideal profile for a capital-intensive infra buildout (knows the asset and the capital markets). Under Blackstone, effectively a founder-CEO operating a sponsor-owned platform: high autonomy on growth, but the monetisation strategy (REIT carve-out, eventual full exit) is Blackstone's call, not his.

Sponsor as de-facto management: Blackstone (Jon Gray's digital-infra franchise) sets capital-allocation and exit strategy. Their QTS track record (+900%) is the single best evidence the platform will be well-stewarded and aggressively grown.

Lens 10 · Forensic Red Flags + Regulatory

Accounting/forensic (bounded by privacy — no audited statements public). As a private with no SEC filings, standard forensic screens (accruals, receivables-vs-revenue, SBC add-backs, segment gaming) cannot be run — n/a — private, no audited financials public. Structural flags to watch when the REIT prospectus finally discloses audited numbers:

• Which EBITDA base anchors the distributable income (the ~US$1B vs ~A$210–220M gap from Lens 5) — this determines whether the REIT yield is real or engineered.
• Development-asset accounting — how much value sits in un-leased pipeline marked at cost vs. fair value; capitalised interest during construction.
• Related-party structure — the REIT will lease/buy assets from the Blackstone-owned private platform; the transfer pricing and the sponsor's retained economics (management fees, ROFR pipeline) are the classic sponsor-REIT conflict to scrutinise.
• Lease-quality disclosure — tenant names, WALE (weighted-average lease expiry), and customer concentration (esp. ByteDance %) — Chindata, a structural comp, ran 86% ByteDance revenue; if AirTrunk's REIT assets carry similar single-tenant skew, that's a material risk the prospectus must quantify.

Regulatory findings (required sub-section):

• SEC (EDGAR LR + AAER): zero — AirTrunk has no CIK and is not an SEC registrant. Verified via regulatory/regulatory-findings.md (generated 2026-07-06).
• Non-SEC enforcement (web search "AirTrunk" (FTC/DOJ/consent decree/settlement/fine/penalty)): no material enforcement actions, fines, or consent decrees found against AirTrunk as of 2026-07-06.
• Sectoral/regulatory risk (not enforcement, but material to the thesis):
• Malaysia/Johor: informal DC moratorium since ~mid-2024; up to ~30% of applications rejected; dedicated DC water tariff (RM5.33/m³) and new power tariffs. Directly constrains AirTrunk's >700 MW Johor pipeline. A citizen petition for a formal moratorium is live.
• Singapore: structural capacity rationing (the 2019–22 pause); AirTrunk holds a scarce pilot allocation but growth is capped by policy.
• ByteDance/US-China: as a TikTok/ByteDance landlord, AirTrunk carries indirect exposure to US restrictions on ByteDance — a demand-side tail risk if ByteDance's APAC footprint is forced to contract.
• Conclusion: No material regulatory or legal enforcement findings — verified via SEC EDGAR EFTS (LR, AAER: no CIK), web enforcement search, and (no 10-K exists) as of 2026-07-06. The regulatory story is jurisdictional resource-constraint risk (power/water), not enforcement/legal risk.

---

Phase D — Project & stress-test

Lens 11 · IPO-Readiness & Path-to-Tradeable (private overlay — the be-early payoff lens)

This is the headline for a MenFem private-frontier reader: AirTrunk is on the cusp of a partial tradeable event.

• Stage: pre-IPO / secondary-active → S-1-imminent (equivalent). On the readiness scale (private-watch.json), this is a 4→5: bank mandate awarded (Citi/DBS/Jefferies, Apr 2026), cornerstone solicitation underway, launch penciled ~Aug 2026, listing ~Sep 2026.
• The instrument: a Singapore-listed REIT ("S-REIT"), raising ~S$1.0–1.5B, trust value ~S$2.5B — potentially Singapore's largest-ever REIT IPO.
• CRITICAL STRUCTURE CAVEAT (the whole investment case pivots on this): the REIT is NOT AirTrunk. It is a carve-out of a stabilised sub-set of assets (~S$2.5B trust vs. the >A$24B / US$30B-committed platform). Blackstone/CPPIB retain the private development engine and drop stabilised assets into the REIT over time (the QTS/Digital-Realty sponsor-REIT model). So a public investor gets a yield vehicle levered to a slice of mature APAC DC cashflows — not the compounding growth of the buildout. The growth (India 5 GW, Johor, Osaka) stays private, captured by Blackstone's funds.
• Milestones that unlock the S-1 / de-risk the listing: (1) prospectus disclosure of the distributable-income base + WALE + tenant concentration; (2) cornerstone commitments locked; (3) constructive Singapore-REIT tape (Keppel DC REIT / Digital Core REIT yields); (4) macro/rates window holding through Sep 2026.
• Estimated window to tradeable: ~2–3 months (Sep 2026) for the REIT sleeve. Full-platform tradeability (a whole-company IPO or trade sale of Blackstone's stake) is a multi-year, later-cycle event — the REIT is Blackstone starting to monetise, not exiting.
• Write-back: AirTrunk should be added to research/private-watch.json (beat: datacenters, stage: pre-ipo, ipo_readiness: 4, lead_investors: "Blackstone, CPP Investments", catalyst: "Singapore S-REIT IPO ~Aug–Sep 2026 (~S$1.5B raise, ~S$2.5B trust) — stabilised-asset sleeve; growth platform stays private", dossier: this file). (Flagged for Connor — not written here, per wave boundaries: this loop does not edit watchlist/state files.)

Forecast (not an EPS line — a binary catalyst, ``): AirTrunk's Singapore REIT prices and lists by 2026-12-31 — I'd put this at ~70% (p=0.70), conditioned on the Singapore-REIT window and rates staying benign; the platform, sponsors, and mandate are all in place, so the residual risk is market-window, not readiness. Per --watchlist rules I do NOT create a forecast.ts entry in this loop — logged here as a note for a future /thesis//deep-dive pass to formalise.

Lens 12 · Bull vs Bear

Bull case. AirTrunk is the single best-positioned pure-play landlord of the APAC AI-datacentre supercycle. APAC DC capacity roughly doubles by 2030 (one source: 32→57 GW, ~12% CAGR; another: →24 GW) on ~US$50B+ of SEA/India pipeline capex. AirTrunk owns the scarce inputs — entitled land + queued power — in exactly the metros absorbing the spillover (Johor, Mumbai, Tokyo). Its contracts are long-dated, inflation-linked, blue-chip — bond-like cashflows on a call option on AI. Under Blackstone it has best-in-class capital access (>A$18B green debt) and a proven sponsor playbook (QTS +900%). The REIT gives a low-risk yield entry and validates the private marks. The pre-mortem's opposite: if power gets built and AI demand holds, this compounds for a decade.

Bear case (permanent-impairment risks).

1. Hyperscaler self-build + lease-flexibility shift. The customers are the competitors. In 2026 they explicitly favour flexibility over long-term lock-in and are self-building ~60%+ of capacity. If the mix tilts further to owned + short leases, AirTrunk's long-dated take-or-pay premium erodes — the entire cashflow-visibility thesis (and the multiple) compresses.
2. Power/water = a hard physical ceiling. Johor rejecting ~30% of applications, water-tariffed; Singapore rationed; grid queues everywhere. Signed leases on un-energisable land = stranded capex and slipped revenue. The buildout is gated by grids AirTrunk doesn't control.
3. Customer concentration (ByteDance). If AirTrunk's leasable base skews to a few tenants (Chindata = 86% ByteDance) and ByteDance retrenches on US pressure, a single counterparty shift is material.

Are the marks too high? ~20–23× EV/EBITDA is a full, AI-peak multiple. It's defensible on secular growth + scarcity, but it prices in flawless execution of a US$30B+ buildout. Any combination of AI-capex digestion + power delays + rate back-up would re-rate it down hard.

Contrarian view (what the market is refusing to see). The consensus treats "AirTrunk IPO" as "buy the APAC AI-datacentre growth story." The market is missing that the REIT is deliberately the low-growth slice — Blackstone is selling the public a stabilised-yield sleeve at the top of the cycle while keeping the compounding development platform private. The public vehicle and the real growth engine are different animals. The genuinely contrarian read: the most interesting thing about AirTrunk (the private buildout) is the one thing you can't buy — and the thing you can buy (the REIT) is being floated precisely because the sponsor wants liquidity near a peak. Buy the REIT for yield if the prospectus numbers hold; do not buy it as a growth proxy.

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

Dismantling the bull case:

• The cashflow-visibility story is more fragile than it looks. "10–15 year take-or-pay" assumes hyperscalers honour long leases. 2026 evidence says they're cancelling early-stage leases (~200 MW pulled in the US), demanding shorter terms, and accepting price variability for delivery flexibility. If that behaviour reaches APAC, AirTrunk's WALE shortens and its "bond-like" premium is a mirage priced at 20×+.
• Revenue concentration is the kill-shot vector. Undisclosed, but structural comps (Chindata 86% ByteDance) and AirTrunk's confirmed ByteDance-anchor status in Malaysia/Singapore suggest dangerous single-tenant skew. US action against ByteDance/TikTok is a live, bipartisan, multi-year threat — a forced APAC contraction would hit AirTrunk's most-leased region.
• The moat is weaker than bulls think on the demand side. AirTrunk has no pricing power over customers (they're 100–1000× its size and self-build) and no power over suppliers (price-taker on the one scarce input, grid power). Its moat is "we built cheap capacity in scarce metros first" — durable for installed assets, but it does not stop a hyperscaler from self-building the next 3 GW.
• Most dangerous competitor bulls underestimate: the customers themselves. Not NEXTDC or GDS — AWS/Microsoft/Google/ByteDance building their own APAC capacity. Every self-built megawatt is a lease AirTrunk didn't sign.
• Worst capital-allocation risk: the US$30B India land-grab. Announcing US$30B/5 GW and a ~US$21B single Mumbai campus — "planned, not committed" — into India's power-and-permitting-constrained grid is the kind of top-of-cycle over-reach that, if AI-capex digests, leaves a stranded pipeline. It reads as sponsor-driven AUM/growth optics as much as disciplined return-on-capital.
• Sponsor-REIT conflict (the structural short on the public vehicle). The REIT buys/leases assets from its Blackstone parent, pays the sponsor fees, and gets fed pipeline at the sponsor's discretion. Public unitholders are structurally junior to Blackstone's economics. Floated near an AI-infra peak, this is a liquidity event for the seller.
• What must hold for today's price: AI-capex keeps compounding without digestion; hyperscalers keep signing long leases; power/water constraints ease enough to energise the pipeline; rates stay benign for a 20×+ infra multiple. Break any one and the equity re-rates.
• Single scenario that permanently impairs: a 2001-telecom-style overbuild — the industry (hyperscalers self-building + every colo racing to add capacity into an AI-capex digestion year) creates a supply glut just as demand growth cools; lease rates fall, un-leased pipeline strands, and a 20×-marked, debt-heavy platform de-rates violently. Plausibility: moderate. Not a 2026 base case (demand is still red-hot), but the precise risk a top-of-cycle, debt-funded, US$30B-pipeline land-grab is most exposed to.

Lens 14 · Management Questions (ordered by information value)

1. What EBITDA base and WALE will the Singapore REIT's distributable income be struck on — and does it reconcile the ~US$1B run-rate vs. ~A$210–220M current-year figures? (This single answer sets the whole REIT valuation.)
2. What is customer concentration across the REIT's assets — specifically, what % of contracted revenue is ByteDance, and what is the largest single tenant?
3. How much of the >1.4 GW is on true long-dated take-or-pay vs. shorter/flexible terms — and has any customer renegotiated toward flexibility in the last 12 months?
4. Which assets go into the REIT vs. stay in the private platform, and what are the exact sponsor economics (fees, ROFR pipeline, transfer pricing)?
5. Of the US$30B India / 5 GW plan, how much is committed capital vs. aspiration — and what secured power do you have for the 3 GW Raigad campus?
6. What is your firm, energised, grid-connected power position by market — and how much signed-lease capacity is currently un-energisable due to grid queues?
7. How exposed is the Malaysian pipeline to the Johor moratorium/water-tariff regime, and how many approvals are pending vs. rejected?
8. If a major hyperscaler cut its APAC leasing 20–30% (US ByteDance action, or AI-capex digestion), what's the revenue and covenant impact?
9. What is net debt / EBITDA across the platform, and how much development capex is funded vs. still to be raised?
10. What share of new demand is AI-training (lumpy, power-dense, liquid-cooled) vs. cloud (steady) — and how does that change your asset design and cancellation risk?
11. How do you compete for the next tranche of hyperscaler capacity against those same customers' self-build programmes?
12. What are the renewal economics on your earliest-expiring leases — pricing power up or down at renewal?
13. What's the long-run exit for Blackstone/CPPIB — REIT drop-downs over time, full IPO, or trade sale — and on what horizon?
14. How much un-leased (speculative) pipeline are you carrying, and at what point does building ahead of demand become a balance-sheet risk?
15. What single scenario keeps you up at night as the thing that could impair the platform — and how are you hedged against it?

---