Eight Percent and Climbing: A Field Guide to Singapore-Listed Data Center REITs
Singapore-listed data center REITs are offering yields that look almost too good in a world starved for income. We dig into the portfolio fundamentals, AI-driven re-rating thesis, and the risks income investors are quietly underpricing.
There is a version of this story that writes itself. AI demand is insatiable, APAC is underserved, Singapore is the gateway to Southeast Asia, and someone is dangling an eight percent yield in your face while global bond markets are still figuring out which way is up. What's not to like?
The honest version is messier. Singapore-listed data center REITs have been through a brutal few years of rate-driven compression, sponsor drama, and some genuinely uncomfortable asset quality questions. The AI tailwind is real. So is the leverage. So are the currency mismatches between Singapore dollar distributions and assets generating ringgit, won, or sterling. Income investors who go in treating these as straightforward bond proxies are going to have a bad time. The ones who do the portfolio work first have a genuinely interesting opportunity.
We've been tracking the underlying assets across APAC closely enough to have opinions. Here is what we actually think.
Why Singapore Became the Listing Hub for Digital Infrastructure Income
Singapore's position as the home exchange for data center REITs is not accidental. The SGX regulatory framework, the maturity of its institutional investor base, and the city-state's own role as a critical interconnection hub in Southeast Asia all converged to make it the natural listing venue for managers who wanted to package digital infrastructure cash flows into distributable income. The four Singapore-listed data center REITs collectively hold a portfolio valued at S$32.5 billion, with a combined market capitalisation of approximately S$24 billion as of 2025. Dr Wealth
The structure that emerged, denominated in Singapore dollars and governed under the Monetary Authority of Singapore's REIT framework, gave managers access to a deep pool of yield-hungry capital from across Asia. For the underlying assets, which were often located in markets like Australia, Europe, or South Korea, the Singapore listing provided valuation credibility and liquidity that local exchanges might not have offered.
The two names that define this category are Keppel DC REIT and Digital Core REIT. They share the SGX listing and the digital infrastructure mandate, but their portfolios, sponsor relationships, and risk profiles are different enough that treating them as interchangeable is a mistake most investors make exactly once.
Keppel DC REIT: The Incumbent With a Complicated Inheritance
Keppel DC REIT listed in 2014 and spent most of its first decade compounding quietly and confidently. It was APAC's first data center REIT, it had a strong sponsor in Keppel Corporation, and its portfolio of colocation facilities across Singapore, Australia, Europe, and eventually China seemed to combine defensive income with genuine growth optionality. For a long time, it was the go-to answer when someone asked what a well-run digital infrastructure REIT looked like.
Then 2022 and 2023 happened. A combination of rising interest rates, a messy situation involving a Chinese client that stopped paying rent, and broader market skepticism about the sustainability of tech-adjacent valuations pushed the unit price down sharply from its peak. The yield, which had looked modest at the highs, started looking attractive again. The question was whether it was attractive because the fundamentals were strong or because the market was telling you something about asset quality.
“The yield went from looking modest to looking interesting to looking almost suspicious, all in about eighteen months. That's the market asking a question. The job is figuring out whether the answer is 'undervalued' or 'value trap.”
Our read on the portfolio is nuanced. The core assets in Singapore are genuinely high quality. Singapore remains one of the most constrained data center markets in the world, with the government's moratorium on new capacity having only recently relaxed, and the colocation assets there generate some of the most defensible cash flows in the REIT's portfolio. Singapore is not just a market where demand exists; it is a market where supply has been artificially throttled, which is about as good as it gets for an incumbent with existing lettable area.
The European assets, particularly in the Netherlands, Germany, and the UK, are also solid. European enterprise and hyperscale demand has been accelerating faster than most forecast, and the assets Keppel DC holds in those markets benefit from both contractual escalators and genuine scarcity value in markets that are running out of power-enabled land.
The China exposure is where the story gets complicated. The Guangdong assets were the source of the client payment crisis, and while Keppel DC has worked through most of the acute issues, the longer term question of how to treat Chinese data center assets in a portfolio that trades to Singapore-based yield investors is unresolved. China revenue is harder to underwrite, harder to repatriate predictably, and harder to value in a framework built around stable distributable income. The market has been discounting it heavily. Whether that discount is adequate is a genuine debate.
On the AI re-rating thesis, Keppel DC sits in an interesting position. Its colocation model, where it operates the facility and leases space and power to multiple tenants, is not the same as the hyperscale wholesale model that has captured most of the AI capital expenditure narrative. But enterprise AI workloads do flow through colocation facilities, and the densification of racks that AI-optimized servers require is already pushing revenue per square meter higher across the portfolio. The question is how much of that densification benefit accrues to the REIT versus the tenants who are retrofitting their own kit.
Digital Core REIT: The Hyperscale Pure Play With Rate Scars
Digital Core REIT listed in late 2021, which in retrospect was about as unlucky a time to list a highly leveraged rate-sensitive vehicle as it is possible to choose. The offering was oversubscribed. The unit price peaked almost immediately and then spent the next two years being methodically repriced as the Fed raised rates twenty times and the market reassessed the cost of capital for long duration income assets.
The portfolio today spans 11 data centers across roughly 1.2 million square feet of net rentable area, with assets under management of $1.8 billion and portfolio occupancy running at 97% as of 2025. Digital Core REIT Those are strong operational numbers. The financial structure that sits beneath them is where the complexity lives.
The portfolio is different from Keppel DC's in ways that matter. Digital Core is more concentrated in hyperscale wholesale facilities, the large-scale campuses that serve the major cloud providers. Its geographic footprint runs across North America, Europe, Japan, and Singapore, and its tenant list reads like a who's who of the hyperscale universe. The underlying sponsor is Digital Realty, the US-listed data center REIT, which provides both the acquisition pipeline and the operational expertise that a relatively small Singapore-listed vehicle could not generate organically.
The AI angle for Digital Core is more direct than for Keppel DC. Hyperscale cloud providers are the ones building the GPU clusters. They are the anchor tenants in Digital Core's facilities. When Microsoft or AWS expands its AI infrastructure, that demand flows into exactly the kind of large campus wholesale arrangements that make up Digital Core's portfolio. The thesis is straightforward: AI capital expenditure becomes hyperscale cloud expansion becomes wholesale leasing demand becomes higher occupancy and better lease economics at renewal.
The complication is leverage. Digital Core listed with a gearing ratio that was already at the high end of what Singapore REIT investors are comfortable with, and the rate environment that followed penalized that leverage hard. Interest coverage ratios compressed, distribution per unit fell from its initial projections, and the unit price reflected all of that. The recovery narrative since late 2023 has been partly about rates stabilizing and partly about a genuine improvement in the leasing environment as AI-driven demand absorbed available capacity faster than expected.
The sponsor relationship with Digital Realty is a double-edged factor. On one hand, Digital Realty is one of the most sophisticated data center operators in the world, with a global development pipeline and a first right of refusal structure that gives Digital Core REIT a path to accretive acquisitions. On the other hand, sponsor-to-REIT acquisition dynamics in Singapore have not always worked cleanly in unitholders' favor, and the potential for related party transactions to prioritize sponsor interests over REIT investors is a structural risk that governance-focused investors take seriously.
What Eight Percent Actually Means
The yield math on Singapore data center REITs requires some unpacking, because the headline number is doing a lot of work.
| Metric | Keppel DC REIT | Digital Core REIT | Context |
|---|---|---|---|
| Structure | Colocation focused | Hyperscale wholesale | Different risk profiles |
| Geographic Core | Singapore, Europe, China | North America, Europe, Japan, Singapore | Concentration differs significantly |
| Lease Profile | Shorter, multi-tenant | Longer, hyperscale anchors | Colocation renews more frequently |
| Sponsor | Keppel Corporation (Singapore) | Digital Realty (US) | Acquisition pipeline source |
| Gearing Sensitivity | Moderate | Higher | Rate cycle impact varies |
| AI Exposure Type | Enterprise colocation densification | Hyperscale cloud buildout | Both benefit, different mechanisms |
| 1Y Price Return (Jul '24–Jul '25) | +15.23% | -30.92% | Source: Dr Wealth |
When yield-hungry investors see eight percent from a Singapore-listed vehicle, the first question to ask is whether that distribution is being supported by actual free cash flow or by financial engineering. For data center REITs specifically, the capital expenditure cycle is lumpy and significant. Maintaining existing facilities, expanding power capacity, and upgrading cooling infrastructure to handle higher density AI workloads all require cash that does not show up neatly in distributable income calculations. If a manager is deferring necessary capital expenditure to protect the distribution in the short term, the yield is not what it appears to be.
The second question is currency. Both Keppel DC REIT and Digital Core REIT distribute in Singapore dollars, but neither earns all of its income in Singapore dollars. Keppel DC has Australian dollar, euro, sterling, and renminbi cash flows that all run through a hedging program before reaching Singapore dollar distributions. Digital Core has US dollar cash flows from its North American assets. Currency hedging costs real money, and in an environment of elevated short term rates and SGD strength, those costs can be meaningfully dilutive.
The third question, which is the one that distinguishes sophisticated analysis from yield-chasing, is what the distribution looks like in three years. Data center REITs have genuine growth levers: occupancy expansion, rental rate escalation, power densification, and acquisition-driven asset base growth. But they also have offsetting pressures: lease expiry risk, the cost of upgrading older facilities, and the interest expense on the debt that funded past acquisitions. A yield that looks stable on a trailing basis can look very different when you model out the next capital expenditure cycle and a refinancing event.
The AI Re-Rating Thesis: What Is Priced In and What Is Not
The AI infrastructure supercycle narrative has been used to justify a lot of things in capital markets over the past two years, some of them deserved and some of them not. For Singapore-listed data center REITs, the question is specific: does the AI demand surge actually translate into better economics for the assets these REITs hold, and if so, how much of that improvement is already in the unit price?
Our view is that the thesis is directionally correct but the timeline and mechanism are more complicated than the bulls typically acknowledge. AI workloads do require data center capacity. That capacity is in real supply constraint across most developed markets. Singapore, as we noted, has been supply constrained by policy for years. The constraint is real, and it does give pricing power to incumbent operators.
Global data center capacity is forecast to reach approximately 200 GW by 2030, up from roughly 100 GW today, implying around 100 GW of new capacity additions over the next five years at a 14% CAGR, and the real estate value creation embedded in that buildout is estimated at $1.2 trillion. JLL The construction cost alone is stiffening: average global data center construction costs rose from $10.7 million per MW in 2025 to a forecast $11.3 million per MW in 2026 JLL, which raises the replacement cost bar for existing assets and provides a structural floor for valuations.
But the translation from AI demand to REIT distributable income has several friction points. First, the most valuable AI infrastructure, the GPU clusters and the high power density compute campuses, is being built by hyperscalers who often prefer to own rather than lease, or who want purpose-built wholesale arrangements that are quite different from the legacy colocation stock that makes up a large part of the Keppel DC portfolio. The demand is real, but it does not always flow to the existing assets in existing REIT portfolios.
Second, the power density requirements of AI infrastructure are genuinely challenging for older facilities. A colocation data center designed for twenty to thirty kilowatts per rack is not straightforwardly retrofitted to handle eighty to one hundred kilowatts per rack. The capital expenditure required to upgrade cooling and power infrastructure is significant, and in a REIT structure where distributions are prioritized, finding the capital for that upgrade cycle without diluting existing unitholders requires careful balance sheet management.
Third, the hyperscale tenants who anchor the most AI-relevant facilities are sophisticated counterparties with significant negotiating leverage. The narrative that AI demand will automatically push up rental rates assumes that tenants cannot simply threaten to build their own capacity or move to a competing facility. In markets with genuine supply constraints, that leverage is limited. In markets where new supply is being developed aggressively, it is not.
1-Year Unit Price Performance: SGX-Listed Data Center REITs
July 2024 to July 2025
Source: Dr Wealth (2025). Returns represent unit price performance only and do not include distributions.
Peer Comparison: How Singapore REITs Stack Up Against Global Alternatives
Income investors evaluating Singapore data center REITs are not operating in a vacuum. The global digital infrastructure income space includes US-listed REITs like Equinix and Digital Realty, Australian-listed operators, and a growing set of private infrastructure funds targeting the same underlying asset class.
Singapore data center S-REITs delivered total returns in the range of 8.5% to 23% in 2024 The Business Times (citing SGX), outperforming most other S-REIT subsectors, though the sector index had given back approximately 2% year-to-date into 2026 The Straits Times, a reminder that the re-rating is not a straight line.
The US-listed peers trade at meaningfully lower yields than their Singapore counterparts, typically in the three to five percent range for the established names. That yield compression reflects the greater liquidity and depth of the US REIT market, the stronger governance track record of the major US-listed vehicles, and the fact that US data center REITs have compounded asset value and distributions at rates that justify lower initial yields. When Singapore-listed vehicles trade at a meaningful yield premium to US peers with comparable underlying asset quality, that premium requires explanation. Sometimes the explanation is genuine risk differences (leverage, currency, sponsor dynamics). Sometimes the explanation is that the market is undervaluing the Singapore vehicle. Distinguishing between those two cases is the core analytical task.
Australian digital infrastructure, represented by names like NextDC and AirTrunk (prior to its Blackstone acquisition), offers a different angle. Australian assets have the advantage of AUD-denominated cash flows for Australian investors, a deep enterprise market, and strong government-driven demand from public sector digitization. But the Australian digital infrastructure sector has historically been structured as operating companies rather than REITs, which means different tax treatment, different capital allocation flexibility, and different income profiles. The REIT wrapper that Singapore provides is specifically designed for income distribution, not growth reinvestment.
Key Players in Singapore-Listed Digital Infrastructure
Asia-Pacific's first data center REIT, focused on colocation assets across Singapore, Europe, and APAC.
Hyperscale wholesale data center REIT sponsored by Digital Realty, with assets across North America, Europe, and APAC.
The private infrastructure market provides a useful valuation anchor. When Blackstone acquired AirTrunk in 2024 at a valuation that implied metrics well above what Singapore-listed REITs were trading at, it confirmed that private capital sees the underlying asset class as genuinely valuable. The gap between public market REIT valuations and private market transaction values is either an opportunity (the public vehicles are cheap) or a warning (private buyers are paying up for control premiums and development optionality that a listed vehicle cannot easily replicate). Both explanations have merit, and the right answer probably contains elements of both.
What Income Investors Are Underpricing
The risks in Singapore data center REITs are not hidden. They are disclosed in every prospectus and results presentation. The question is whether investors are weighting them correctly.
Interest rate sensitivity remains underappreciated in a market that has been conditioned by two years of "rates will come down soon" expectations. Both Keppel DC REIT and Digital Core REIT carry gearing levels that make them meaningfully sensitive to refinancing costs. If the rate environment stays higher for longer, the squeeze on interest coverage ratios is not just an earnings headwind; it becomes a constraint on the manager's ability to maintain current distribution levels while also investing in portfolio maintenance and growth.
Lease concentration risk is the other factor that gets glossed over in the AI demand excitement. A small number of large tenants in any data center REIT portfolio means that a single non-renewal decision, particularly in a wholesale or large enterprise lease context, can have an outsized impact on income. The probability of any individual lease not renewing is low. The impact if it happens is high. That asymmetry deserves more weight than a simple "high occupancy" headline suggests. Digital Core REIT's 97% portfolio occupancy is a strong number today. It also means there is almost no cushion if a major tenant makes a different decision at renewal.
The governance structure of Singapore REITs, where the manager is a separate entity with its own incentive structure and fee schedule, creates a persistent tension between manager interests and unitholder interests. Acquisition fees, performance fees, and the general incentive to grow assets under management can lead to decisions that benefit the manager more than the investors providing the capital. This is not a Singapore-specific problem; it exists in the US REIT market too. But Singapore's REIT regulatory framework has historically given managers more latitude in related party transactions than the most stringent governance frameworks would prefer.
The Growth Runway: What Actually Has to Happen
For the bull case to work across a five year horizon, several things need to go right simultaneously. AI demand needs to sustain its current trajectory and continue flowing into data center capacity rather than being absorbed by efficiency improvements in the models themselves. Interest rates need to normalize enough to reduce refinancing pressure and potentially re-rate the sector back toward the multiples it commanded in 2020 and 2021. The Singapore government's easing of its data center moratorium needs to translate into Keppel DC and other incumbents being able to expand their Singapore capacity, which has historically been the most valuable part of their portfolios. And the acquisitions that both vehicles execute need to be accretive on a cost of capital basis, not just accretive to assets under management.
None of these are unreasonable expectations. But they all need to go right at roughly the same time, and in a world where geopolitical complexity around technology supply chains, data sovereignty regulations across Southeast Asia, and the general uncertainty of the AI hardware cycle are live variables, "multiple things going right simultaneously" is a higher bar than it sounds.
The more conservative bull case, which we find more credible, is simply that these are well-located assets in structurally supply-constrained markets, the yield is real and backed by contracted cash flows, and the AI re-rating is a genuine but gradual tailwind rather than an immediate revaluation event. The broader structural case is hard to dismiss: JLL forecasts approximately $1.2 trillion in real estate asset value creation from new data center capacity between 2025 and 2030, with global construction costs now exceeding $11 million per MW, a replacement cost floor that benefits existing REIT portfolios. JLL On that basis, Singapore-listed data center REITs offer income investors a combination of yield, digital infrastructure exposure, and APAC geographic diversification that is genuinely difficult to replicate through other listed vehicles.
That is an interesting proposition. It is not a simple one. The investors who will do best here are the ones who have done the portfolio work, understand the leverage and currency dynamics, and are buying the income stream rather than the AI hype cycle.
The yield is eight percent and climbing. Whether it stays there, compresses as the sector re-rates, or turns out to have been the market's way of telling you something you should have listened to, depends almost entirely on which version of this story you believe and what the underlying assets actually do over the next several years.
We know where the facilities are. We know what they hold. The financial story is yours to complete.