This page is still being completed: the data is in, the prose modules are not yet finished and cleared for publication, and the page is set to noindex.
Opportunity rank #30 (cash-flow board) · Singapore (listing venue); assets: Keppel ≈62.8% in Singapore with the rest across APAC and Europe, Digital Core entirely in the US/Canada/Germany/Japan

Singapore Data Centre REITs: Keppel's Premium vs Digital Core's Discount

Our take: Do, with conditions
5–7.2% Distribution yield (two DC REITs) · Verified range · Data as of: 2026-11-09

Data updated: 2026-11-09

Proposed rank 30 (in research). SGX has exactly two pure-play data-centre REITs, and they are mirror images. Keppel DC REIT: a ≈5.05% distribution yield (TTM at the 9 Nov 2026 price), a record FY2025 DPU of 10.381 cents, and 62.8% of assets in Singapore — at the cost of a ≈27% premium to NAV, a single tenant at 43.5% of rent, and roughly a fifth of the DPU paid as capital distribution. Digital Core REIT: the highest DC yield on SGX at ≈7.1% (USD), yet a ≈37% discount to NAV — because it has been hit head-on by two tenant bankruptcies (Sungard 2022; Cyxtera 2023, at 22.4% of income), both bridged by its sponsor, leaving the DPU permanently stepped down by about a tenth and a payout ratio of 100% of income with no buffer. Why local assets are so scarce: Singapore paused new data centres in 2019 and has formally released only ≈280MW over seven years through pilots and the green roadmap, with PUE and green-power thresholds tightening each round; Keppel's Singapore portfolio is a pre-pause legacy injection that later entrants cannot replicate. The verdict is 'do, conditionally': Keppel only as the core leg and not above a 30% premium; Digital Core only as a high-risk satellite sized for another large-tenant event; neither replaces the rank-7 broad S-REIT core. The entry also settles a structural question — why the rank-71 US data-centre REITs are 'growth stocks, do not buy for cash flow' while this one can be bought: same asset class, but 5.0–7.2% distributions with no dividend withholding in the Singapore structure, versus 2.1–2.8% minus 30% withholding held directly in the US.

In short: four lines first

What it costs to get in

Low (listed SGX units, from one lot; at current prices a few hundred Singapore dollars to start — position size is the investor's choice)

Roughly what it earns

5–7.2%(Distribution yield (two DC REITs))

How much work for you

Strong (listed REITs, no operations; quarterly monitoring of leverage, fundraisings and tenant events only)

The biggest catch

Keppel's single-tenant concentration: the largest tenant is 43.5% of rental income (June 2026; 38.2% pro forma after the Tokyo acquisition), and the top ten are roughly 80%. Large leases and few tenants are the data-centre model's natural shape — but at 43.5%, one tenant's renewal negotiation can rewrite the distribution table.

The numbers that matter

Distribution yield (two DC REITs)
5–7.2%
Verified range As of: 2026-11-09 Source: SGX Research: Data Centre REIT sector report (Syfe, Apr 2026)
Keppel DC premium to NAV
27%
Operator estimate As of: 2026-11-09 Source: Keppel DC REIT Annual Report 2025
Digital Core discount to NAV
-39–-36%
Verified range As of: 2026-11-09 Source: SGX Research: Data Centre REIT sector report (Syfe, Apr 2026)
Keppel DC largest tenant share of rent
43.5%
Verified As of: 2026-11-09 Source: Keppel DC REIT acquisition announcement (Sep 2026)
Keppel DC Singapore share of AUM
62.8%
Verified As of: 2026-11-09 Source: Keppel DC REIT 1H 2026 results presentation
Digital Core largest tenant share of rent
30%
Verified range As of: 2026-11-09 Source: Digital Core REIT 1H 2026 interim financial statements (SGX)

Why we think it makes money

The distributions are announcement-grade cash, not model output. Both REITs' payouts, leverage and occupancy come from SGX announcements and annual reports: Keppel's FY2025 DPU is 10.381 cents, up 9.8%, and its DPU is computed after capital-expenditure reserves from distributable income — broadly aligned with this board's 'after all trust-level costs' cash-flow definition (the capital-distribution component aside; see costs). Digital Core's DPU has stepped down from 3.98 to 3.60 US cents over three years and flattened — and the reason (tenant bankruptcies) is traceable through its announcements end to end. Operating businesses on this board cannot reach this evidence standard; it is a structural advantage of the REIT form.

Power-cost pass-through is written into the annual report — the single most important line for a data-centre asset. Keppel's 2025 annual report states that electricity costs are mostly passed through to colocation tenants, while master-lease tenants contract directly with power suppliers; the FY2023 quantified figure was over 90%. Singapore power is the region's most expensive, but under this lease structure most of that bill lands on tenants, not the REIT. A small self-operated facility cannot do this — the power bill lands on it in full — which is precisely why small-capital self-operated data centres fail in Singapore while listed DC REITs work.

The scarcity is policy-made, and the timeline is fully documented: the 2019 pause on new data centres → the 2022 pilot DC-CFA (PUE ≤1.3) → just 80MW awarded in July 2023 to Equinix, Microsoft, GDS and an AirTrunk–ByteDance consortium → the 2024 Green Data Centre Roadmap (≥300MW near term, 200MW of it tied to green power) → DC-CFA2 in August 2026 releasing another 200MW (PUE ≤1.25, ≥50% green energy). Seven years, roughly 280MW formally released. Keppel's ≈62.8% Singapore portfolio is pre-pause legacy stock; Digital Core holds no Singapore assets at all (11 properties across the US, Canada, Germany and Japan, with a 2.5% interest in SIN12 only announced in 2026) — for Digital Core, 'Singapore data-centre REIT' is a listing venue, and this page does not dress that up.

The two risks are priced side by side in one entry, ready to be chosen between: Keppel is certainty at a 27% premium — record distributions, interest coverage above 6.5×, Singapore assets underneath, expensive but structurally clean. Digital Core is repair at a ≈37% discount — its 7.1% yield contains the consideration for two bankruptcy events and a 30% single-tenant exposure. The cash-flow investor's job is not to pick a side but to size by price discipline: core money buys certainty, satellite money buys the discount, each with its flip conditions written down.

How to actually get in

  1. Core route (Keppel DC): one discipline — do not chase the premium. Stage purchases only at a premium to NAV of 30% or less; wait otherwise. Its role is a data-centre enhancer inside the rank-7 broad S-REIT core, not a replacement. After buying, check three things each quarter: interest coverage (sliding from 7.5× to 6.5×), the largest tenant's lease progress, and any new placement.
  2. Satellite route (Digital Core): size the position by stress test first — assume the largest tenant (≈30% of rent) fails and DPU steps down by a tenth or more, on the Cyxtera precedent, and only buy what that loss is affordable on. Add only on delivery: the Linton Hall lease starting December 2026 and quarterly data supporting the FY2027 guidance (4.15–4.25 cents) — never pay for guidance alone.
  3. Combined view: an equal-weight pair yields ≈6%, half Singapore-asset certainty and half global-asset discount repair. Division of labour against the rank-7 S-REITs (industrial + retail, 5.5–6.4%): broad base first, DC enhancement second, Digital Core strictly as satellite — with a self-set cap on total data-centre exposure.
  4. Pre-purchase checklist (both): latest-quarter leverage and interest coverage, debt maturing within 12 months, any announced equity raising, and the largest tenant's lease expiry schedule. For Keppel, also track the Guangdong Bluesea provisioning; for Digital Core, the EGM outcome on the SIN12 acquisition (2.5% interest in a Singapore asset) — not verified this round, see the unverified list.

What it costs, what it earns

Entry amount: Low (listed SGX units, from one lot; at current prices a few hundred Singapore dollars to start — position size is the investor's choice)
How much work for you: Strong (listed REITs, no operations; quarterly monitoring of leverage, fundraisings and tenant events only)

Costs & taxes, item by item:

Basis: this site uses full-cost net returns only — income minus vacancy, management, maintenance, taxes and insurance, then financing interest, with entry costs in the denominator. See the methodology。

Identity & tax

Status (residency / citizenship)

新加坡全球投资者计划 GIP(直接 PR)

This asset does not count toward the channel This asset leads to no status

Threshold: S$10,000,000 起(新设/扩充本地企业);基金路线 S$25,000,000

在 SGX 买数据中心 REIT 不给任何身份:新加坡没有被动投资者居留。真正的投资移民只有 GIP——S$1,000 万投入本地企业(直接给 PR)、S$2,500 万入 GIP 精选基金,或单一家办路线;二级市场买 REIT 单位不产生资格。新加坡卖的是规则稳定,不是签证;入籍还要求放弃原国籍(不承认双重国籍)。

Source: 新加坡经济发展局(EDB)《GIP 说明书》;ICA

Tax (investor level)

属地税制 + 无资本利得税 + 无股息预提——S-REIT 分派在新加坡层面的摩擦是全榜最低的一类;Digital Core 的底层资产税在信托结构内处理。

  • 属地税制:个人的境外来源收入即便汇入新加坡也一般免税(第 10L 条针对的是低实质实体,不是个人)。
  • 资本利得:不征(属收入性质的交易利得除外,投资者/交易者界线逐案判断)。
  • 股息预提:无。S-REIT 分派对非居民「非个人」投资者预提 10%(优惠延续至 2030-12-31);个人投资者分派在新加坡层面免税。
  • 币种与底层:Keppel DC 分派以新元计;Digital Core 分派以美元计、资产全在海外(美/加/德/日),所在国税务在信托结构内先行处理,单位持有人面对的仍是新加坡层面的分派口径,另有汇率敞口。
  • 对照线:非居民个人直接持有新加坡房产收租,净租金按 24% 征税——小资本在新加坡拿数据中心敞口,只能走上市通道,这也是本条目存在的结构性理由。

Double-tax treaties: 新加坡有 90 多个全面税收协定,与中国、印度均已生效;与美国没有全面所得税协定。新加坡是 CRS 参与方。

Source: 新加坡税务局(IRAS);2025 年财政预算案

This section is general information only — not immigration or tax advice. Program rules change frequently; before acting, rely on the official texts and a licensed adviser.

Risks: what could go wrong

Keppel's single-tenant concentration: the largest tenant is 43.5% of rental income (June 2026; 38.2% pro forma after the Tokyo acquisition), and the top ten are roughly 80%. Large leases and few tenants are the data-centre model's natural shape — but at 43.5%, one tenant's renewal negotiation can rewrite the distribution table. If these happen, our take changes: If the largest tenant's lease fails to renew, its rent share rises above 50%, or Keppel is forced into a discounted equity issue, this page's verdict on Keppel flips to 'do not do / review for exit'.
Keppel's premium dependence: units in issue are up roughly 59% in two years, and every accretive acquisition rests on issuing at today's ≈27% premium to NAV. The premium is granted by the market, not by contract — if AI/data-centre sentiment cools and the price returns to NAV, the growth engine turns dilutive, and pro-forma leverage of 38.8% leaves limited room against its 40% internal cap. If these happen, our take changes: If Keppel's premium narrows below 10% alongside a discounted placement, or interest coverage falls under 5×, this page's verdict flips to 'watch' pending a re-underwrite of the growth model.
Digital Core's event history is not a tail hypothesis — it has happened twice. Sungard (2022, ≈7.1% of income) was bridged by a sponsor cash-flow support agreement (capped at US$7.5M, and structured as an advance repayable across 2024–2028). Cyxtera (2023, second-largest tenant, 22.4% of income) was resolved by a four-party transaction that sold two Silicon Valley assets, cutting exposure from 22% to ≈5% at the cost of a permanent ≈10% DPU step-down. The sponsor stepped in both times; support has never been free, and there is no commitment it happens a third time. If these happen, our take changes: If any of Digital Core's top-three tenants enters bankruptcy, restructuring or material arrears, this page's verdict on Digital Core flips to 'do not do' immediately, without waiting for a sponsor package.
The policy basis of the scarcity premium: Singapore's installed base is ≈1.4GW (IMDA, including owner-occupied), and only ≈280MW has been formally released in the seven years since the 2019 pause; vacancy is ≈4.8% (Cushman & Wakefield, 1H 2026). Keppel's Singapore portfolio carries this policy scarcity premium in its valuation — a materially looser capacity policy would pull the rug from that logic. If these happen, our take changes: If a future DC-CFA round releases materially more capacity (e.g. over 500MW in one round at visibly looser thresholds), or renewal rents turn negative for either REIT, the scarcity-premium logic inverts and this page's verdict drops to 'watch'.

What we have not verified yet

The following key figures have not yet been verified; the reason is noted for each item.

FAQ

Rank 71 already covers US data-centre REITs — why does this entry exist?
Same asset class, opposite holding structure. The US names (Equinix, Digital Realty) yield just 2.1–2.8% and foreign holders face 30% US dividend withholding — they fail as cash-flow purchases, which is why that entry sits at rank 71. The two Singapore listings yield 5.0–7.2% with no dividend withholding at the Singapore level and tax-exempt distributions for individuals — a full order of magnitude less friction. The difference is entirely structure and tax.
Is Keppel's 27% premium too expensive?
The premium buys two things: policy-manufactured scarcity of core Singapore DC assets (which cannot be replicated under the quota regime — see the policy section) and a decade-long distribution record. But be clear-eyed: its growth model issues new units above NAV to buy assets accretively, so the premium is the foundation of the model — if it disappears, accretive growth stalls and discounted issuance becomes possible. Hence the discipline: a hard premium cap, staged buying, and awareness that roughly a fifth of the DPU is capital distribution rather than pure rent.
Is Digital Core's 7.1% yield a value trap?
It is risk pricing, not mispricing. All 11 assets are overseas, the largest tenant is ≈30% of rent (top three ≈58%), and two tenant bankruptcies (Sungard 2022; Cyxtera 2023, 22.4% of income) were worked through only with sponsor support — at the cost of a permanent ≈10% step-down in the DPU level. It now pays out 100% of income with no retained buffer and is expanding via leverage (39.2%) rather than equity. At satellite size, 7% is fair compensation for event risk; at core size, it is a trap.
How does this relate to the rank-7 Singapore S-REIT entry?
Rank 7 is the diversified core: industrial and retail S-REITs yielding 5.5–6.4% across sectors. This entry is a single-sector enhancer whose returns concentrate in data centres — with a 43.5% single-tenant exposure in Keppel and a bankruptcy-event history in Digital Core, so volatility and tail risk are both higher. The sensible structure is broad base first, DC enhancement second — not DC instead of broad.

Related opportunities

Data dates are marked per metric; page updated 2026-11-09. This site publishes information and is not investment advice; the corrections channel is in the Disclaimer。