Cloud Computing
PrivateA best-in-class landlord to the AI-inference buildout, not an operator — the +51% Singapore rent-reversion engine is real and DPU is compounding double-digits, but at ~1.4x P/NAV and a ~4.4% yield you are paying a growth multiple for a REIT whose top-10 tenants are ~80% of income and whose China tail is still not cleanly cut. Own the reversion story, not this entry price.
Research
The verdict
A best-in-class landlord to the AI-inference buildout, not an operator — the +51% Singapore rent-reversion engine is real and DPU is compounding double-digits, but at ~1.4x P/NAV and a ~4.4% yield you are paying a growth multiple for a REIT whose top-10 tenants are ~80% of income and whose China tail is still not cleanly cut. Own the reversion story, not this entry price.
The business model in one line: Keppel DC REIT is the pure-play data-centre landlord of the Keppel group — it buys stabilised, income-producing data centres (often from its own sponsor's development pipeline), leases them to hyperscalers and colocation clients on long contracts, and pays out ~the whole net rental stream as a tax-transparent distribution. It was the first pure-play data-centre REIT listed in Asia (SGX IPO December 2014).
Scale (latest disclosed):
Products / lease types (this is the revenue architecture):
Customers / counterparties: global hyperscalers (cloud, internet-enterprise, telco). KDC SGP 7 & 8 are 100% contracted to global hyperscalers on a colocation basis. Management states the majority of rental income is from clients with investment-grade or equivalent credit profiles — a claim the China tenant default (Lens 10) directly qualifies.
Contract structure & key terms: long WALE (6.7 years at end-2025, 6.9 years at 1H2025), with built-in annual rent escalations on the newer hyperscale contracts (e.g. Tokyo DC 3 has annual escalators, "offering greater cash flow resilience compared to the predominantly fixed-rent contracts in the Japanese market"). The economic engine is the spread between in-place rents signed years ago and today's power-scarce market rent at renewal.
Manager / sponsor: externally managed by Keppel DC REIT Management Pte. Ltd., a subsidiary of Keppel Ltd (via Keppel Capital). Keppel is both sponsor (asset pipeline) and a substantial unitholder — an alignment and a conflict (Lens 9).
For a REIT the "supply chain" is the asset-sourcing and tenancy chain — where the buildings come from, who powers them, and who rents them. Named stakeholders:
Upstream — asset & development pipeline:
Midstream — power, land, cooling (the true chokepoint):
Downstream — tenants → end demand:
Chokepoints (named): (1) Singapore grid power allocation; (2) the sponsor's pipeline + IPT approval gate; (3) tenant concentration (~80% top-10); (4) leasehold land tenure. Names present → lens satisfied.
The moat is location + scarcity + sponsor, in that order — and it is a real but bounded moat.
Bargaining power — who needs whom: In supply-constrained Singapore/Dublin, the landlord holds the whip at renewal (that is the reversion). But against a single hyperscaler that is ~a double-digit % of your income, the tenant holds structural leverage on a new lease or a non-renewal — so the bargaining power is asymmetric by market and by tenant size. The moat is strongest exactly where supply is tightest (SG/Dublin/Japan) and weakest where it isn't (China — where the moat failed outright, Lens 10).
segments.csv is empty (seed-only), so segmentation is ****, not ``. Keppel discloses by geography and by lease type, not by a clean product P&L.
By geography (AUM mix, ~end-2025 / 1Q2026):
By trend (the story the numbers tell):
Latest full-year print — FY2025 (announced 30 Jan 2026) — a strong, largely acquisition-fuelled beat on the growth line:
| Metric (FY2025) | Value | vs FY2024 | Provenance |
|---|---|---|---|
| Gross revenue | S$441.4m | +42.2% | |
| Net property income | S$383.3m | +47.2% | |
| Distributable income | S$268.1m | +55.2% | |
| DPU | 10.381 cents | +9.8% (record) | |
| Aggregate leverage | 35.3% | up from ~17.3% end-2024* | |
| Avg cost of debt | ~3.0% (FY25); 2.8% (4Q25) | lower | |
| Portfolio occupancy | 95.8% | stable | |
| WALE | 6.7 years | ~flat (6.9y at 1H) | |
| Debt headroom to 40% cap | ~S$531m | — |
*The gearing jump end-2024→end-2025 reflects the debt-funded portion of the KDC SGP 7&8 + Tokyo acquisitions; equity was raised too (see below). Note: the ~17.3% end-2024 figure I could not independently source cleanly — flag as n/a at that precision; what is solid is that gearing rose materially into the 30s and stands at 35.3% end-2025.
Interim confirmation — 1H2025 (30 Jun 2025): DPU 5.133 cents (+12.8% YoY), distributable income +57.2% to S$127.1m, NPI +37.8%, gearing 30.0%, occupancy 95.8%, WALE 6.9y, portfolio rental reversion ~+51%. 1Q2026: DPU +13.2% YoY, reversion ~+51% — the double-digit DPU cadence is holding.
What drove it: (1) full-period contribution from KDC SGP 7 & 8 (the S$1.38bn hyperscale acquisition, ~+8.1% DPU accretive) and Tokyo DC 1 & 3; (2) violent positive reversions on Singapore colocation renewals; (3) finance income from vendor notes (the Macquarie AU DC Note and similar structures) — a genuine but lower-quality earnings input than pure rent; (4) falling cost of debt.
Balance-sheet flags: gearing has climbed from the high-teens into the mid-30s in ~18 months — the acquisition sprint consumed the balance-sheet cushion. Still below the 40% MAS-comfortable / internal cap with ~S$531m headroom, but the days of a fortress ~17% gearing are over; further large acquisitions need equity. ~71.2% of borrowings hedged, ~76% fixed, ~3.1–3.3y weighted debt/hedge tenure — reasonably defended against rate moves but with refinancing to roll.
Market reaction / what's priced in: the stock trades at ~1.4x P/NAV and a ~4.4% forward yield — i.e. the market has already capitalised the reversion story into the price. A strong print no longer surprises; the risk is now to expectations, not to the fundamentals (Lens 12).
transcripts/ is empty, so this is ****. Reading the arc of management commentary FY2023 → 1Q2026:
What they stopped saying: the Guangdong problem is no longer front-and-centre in the narrative — which is a tell, because (Lens 10) it is not fully resolved. The tone shift from "loss allowance" (2024) to "record DPU" (2026) is real and earned, but it has also conveniently buried an unresolved credit tail.
| Name | Ticker | Type | Fwd distribution yield | P/NAV | Gearing | Source |
|---|---|---|---|---|---|---|
| Keppel DC REIT | AJBU.SI | Pure-play DC REIT (SG) | ~4.4% (S$2.27, Jun'26) | ~1.4x | 35.3% (end-'25) | |
| Digital Core REIT | DCRU.SI | Pure-play DC REIT (Digital Realty sponsor) | ~8.2% | n/a (trades <1x NAV) | ~low-30s% | |
| Mapletree Industrial Trust | ME8U.SI | Industrial + DC (largest DC AUM S$5.06bn) | ~6.8% | n/a | n/a | |
| NTT DC REIT | (SGX, 2025 IPO) | Pure-play DC REIT (NTT sponsor) | ~8.1% | n/a | n/a | |
| Digital Realty | DLR (NYSE) | Global DC landlord | n/a | n/a | n/a | anchor only |
| Equinix | EQIX (Nasdaq) | Global DC landlord | n/a | n/a | n/a | anchor only |
Read of the table: Keppel DC REIT is the expensive name in its own peer set — it yields ~4.4% while Digital Core, MIT and NTT DC REIT yield 6.8–8.2%, and it trades at ~1.4x NAV while several SG DC peers trade below NAV. The market is explicitly paying up for (a) the best DPU-growth trajectory in the group (+9.8% FY25, +13.2% 1Q26 vs MIT's −6.3%), (b) the Singapore reversion engine, and (c) the Keppel sponsor pipeline + governance quality. The premium is a considered verdict, not an anomaly — but it means the yield cushion that normally protects a REIT investor is thin here. You are underwriting growth, not income.
Pattern of the ~5-year tape (all ``):
What the market actually reacts to for AJBU: (1) tenant/credit events (asymmetric downside — the concentration risk made visible), (2) acquisition + equity-raise mechanics (dilution vs accretion), (3) reversion data points (the bull engine), (4) rates (bond-proxy beta). It reacts less to broad tech/AI-capex headlines than a chip name would — this is a landlord, and its tape trades on rent, credit, and yield.
Structure first (this is an externally managed REIT — the key governance fact). The REIT has no employees; it is run by Keppel DC REIT Management Pte. Ltd., a Keppel Ltd subsidiary. Fees flow to Keppel; the asset pipeline comes from Keppel; and Keppel is a large unitholder. Alignment and conflict are structurally fused.
Red flags (management-specific): (1) the interested-person-transaction dependency — most growth is buying assets from your own sponsor, requiring EGM approvals and independent valuations; (2) external-manager fee drag + potential empire-building incentive (fees scale with AUM, so there is a structural bias to acquire and issue equity); (3) the DPU-smoothing via finance income and capex/ULP reserve mechanics flagged by The Edge on the FY2025 result.
Acting as a forensic analyst. Two real, named credit events sit at the centre of this REIT's recent history — this is where the "investment-grade tenant" story meets reality.
Regulatory findings (required sub-section). Per regulatory/regulatory-findings.md: Keppel DC REIT has no SEC CIK and is not an EDGAR filer, so no SEC Litigation Release or AAER search is possible or applicable (total_sec_findings: 0 — because it is out of SEC jurisdiction, not because it was cleared by the SEC). Web search for non-SEC enforcement ("Keppel DC REIT" (FTC OR DOJ OR consent decree OR settlement OR fine OR penalty)) returns no securities-regulator enforcement action against the REIT — the only "settlements" are commercial disputes with tenants (DXC, Bluesea), not regulator actions. As a Singapore issuer it is regulated by MAS + SGX; no MAS/SGX enforcement finding surfaced. Conclusion: no securities-regulatory or enforcement findings — verified via SEC EDGAR EFTS (N/A, no CIK), web search, and public disclosures as of 2026-07-06. The material risks are credit/accounting, not enforcement:
Guangdong / Bluesea sale-leaseback default (the live wound). The Guangdong DC 1, 2 & 3 assets were acquired as a triple-net sale-leaseback from Guangdong Bluesea Data Development (DC1 in Jul 2021 for ~S$98.1m, 15-year lease). Bluesea defaulted on rent; KDC issued a letter of demand for RMB48.3m (~S$9m) in arrears, of which Bluesea had settled only RMB0.5m. KDC recognised a loss allowance of ~S$5.3m in 1Q2024, a −0.326c / −13% hit to 1Q2024 DPU. Forensic significance: (a) a triple-net structure concentrates all credit risk on one tenant and gives the landlord less operational control, not more; (b) the "majority investment-grade tenants" claim did not protect this asset; (c) as of the latest disclosures the Guangdong assets still sit in the portfolio — the market expected a clean exit that has not clearly happened, and rental income + offsetting loss allowances are still running through Gross Revenue / Property Expense. This is the single most important thing a bull is under-weighting. Watch for either a completed divestment or a full write-down.
DXC Technology default (resolved). DXC partially defaulted on colocation payments at Keppel DC Singapore 1; KDC won in the Singapore High Court (Jan 2024) and reached a S$13.3m GST-inclusive full-and-final settlement (~S$11.2m net, spread across FY24). Cleanly resolved — but the second independent tenant-credit event in the same window underscores that concentration risk is not theoretical for this name.
DPU quality / non-GAAP flattering. FY2025 DPU was struck after setting aside capex and ULP (unrecovered-loss-provision) reserves, and reported distributable income is boosted by finance income from vendor notes (Macquarie AU DC Note, etc.). Neither is improper, but both mean reported DPU is a slightly softer number than pure in-place rent would give — a forensic reader should haircut the headline DPU-growth when comparing to organic rent growth.
Standard REIT accounting watch-items: fair-value revaluation gains flatter NAV in a market where DC cap rates could widen if rates stay high (NAV is mark-to-model); leasehold land amortisation; FX translation across 10 countries into S$; and rising SBC-equivalent (manager fees) as AUM grows. All routine, all / — none currently alarming, but the NAV is model-dependent and the ~1.4x P/NAV magnifies any cap-rate widening.
Net forensic read: the accounting is clean and well-governed; the risk is credit concentration + an unresolved China tail + slightly-flattered DPU, not fraud.
No forecast.ts create in --watchlist mode (per skill). Projection is `` with arithmetic shown; anchored on FY2025 actual DPU 10.381c. DPU (not EPS) is the right unit. Fiscal year = calendar year.
Base case (FY2026E–FY2028E):
Bull path: sustained SG/Dublin power scarcity keeps reversions >30% through 2027, a large accretive sponsor acquisition lands with an equity raise the market rewards, and China is exited clean → FY2028E DPU ~13c+.
Bear path: rates stay higher-for-longer (bond-proxy de-rate + refinancing at higher cost), a top-10 hyperscaler doesn't renew or re-prices down at a soft European asset, China forces a write-down, and the reversion cliff arrives faster than expected → FY2028E DPU flat-to-down vs FY2025 (~9.5–10.4c) and a yield re-rating toward peers (6%+) that would mean a lower unit price even with flat DPU.
The number that matters for a REIT: not just DPU growth but DPU growth vs the yield you buy at. At ~4.4% entry, ~6–7% FY26E DPU growth gives a ~11% total-return math if the multiple holds — but the multiple (1.4x P/NAV) is the fragile input, not the DPU.
Bull case. Keppel DC REIT is the cleanest listed way to own the physical scarcity underneath the AI-inference buildout in Asia's best DC markets. The moat is a regulator-created supply cap (you cannot build the substitute in Singapore), and it is cashing in right now as legacy colocation rents reprice +45–51% into a market with no new supply. DPU is compounding double digits (+9.8% FY25, +13.2% 1Q26) — the best trajectory in the SG REIT complex — funded by a falling ~3.0% cost of debt, a 6.7-year WALE, 95.8% occupancy, and a Keppel sponsor pipeline (~S$3.7bn) that gives visible external growth. Management is deep-pedigree institutional, governance is award-winning, and the portfolio is actively upgrading — shedding China/Malaysia tail, buying AI-ready hyperscale SG + Japan. Contrarian bull kicker: the market treats "AI infrastructure" as a chip/hyperscaler trade and under-owns the landlord that captures the same demand with a fraction of the technology-obsolescence risk and a contracted 6.7-year income stream.
Bear case. You are paying ~1.4x NAV for a ~4.4% yield — a growth multiple on an income vehicle, which inverts the usual REIT safety (thin yield cushion, NAV that is mark-to-model and vulnerable to cap-rate widening if rates stay high). Three things could permanently impair the thesis: (1) tenant concentration — top-10 ≈ 80% of income; a single hyperscaler non-renewal or a soft-market down-reversion (Europe) hits DPU hard, and the tape shows credit events move this stock more than anything else; (2) the reversion is a one-time repricing, not a perpetuity — once legacy rents are marked to market, DPU growth cliffs back to escalator-driven mid-single-digits, and a 1.4x multiple does not survive that; (3) the China tail is not cut — Guangdong/Bluesea is still in the portfolio with an unresolved credit loss the narrative has quietly buried. Pre-mortem (18 months out, thesis broke): rates stayed elevated → the bond-proxy de-rated toward a 6% yield and a European hyperscaler either left or re-priced down and China forced a write-down — so the unit is at ~S$1.80 despite roughly flat DPU, because the multiple compressed. The market is currently refusing to see that almost all the DPU beat is acquisition + one-time reversion, not durable organic growth, and that the premium multiple is the whole risk.
Are multiples too high? For the quality — yes, defensibly; for the risk-adjusted entry — the premium leaves no margin of safety. This is a great asset at a full price.
Dismantling the bull case.
A real, cash-generating neocloud retrofitter trading at ~18x trailing sales on a single $865M Nscale contract and a still-71%-Bit-Digital-controlled cap table — the build is genuine, but the multiple already prices the NC-1 inflection that hasn't happened yet.
A merchant-power balance sheet wearing a regulated-utility's contracted growth — long-dated nuclear PPAs to AWS/Meta de-risk the AI-demand story, but the GAAP P&L is hostage to hedge mark-to-market and the equity carries ~3.4x the net debt of Constellation. Cheapest large-cap way to own the data-center power trade if (and only if) ERCOT/PJM load growth shows up; bull at ~10x forward EBITDA, but leverage + commodity beta make it the high-volatility expression, not the safe one.
The default arms dealer of the AI buildout — a real moat compounding a $15B backlog into 30% organic growth, but priced at 82x for perfection while insiders sell 65:0 and EMEA orders are already cracking.