INTRODUCTION TO DATA CENTRE FINANCING

The surge in data centre financings has been driven primarily by the rapid expansion of data generation and digitisation. As organisations increasingly outsource technology infrastructure to specialised operators to meet growing digital demands, reliance on large-scale data centre capacity has intensified. This trend has been significantly amplified by the rapid increase in computing requirements associated with the rise of artificial intelligence ("AI"). The scale and intensity of AI workloads, particularly for hyperscalers training and operating advanced models, have accelerated demand for high-density computing capacity at a pace that exceeds currently available supply. As such, expanding data centre capacity is no longer optional but imperative. The resulting structural supply-demand imbalance has in turn increased dependence on external financing to fund new developments and capacity expansions. Against this backdrop, investment in data centres is being shaped by several interconnected forces. Spending by major operators continues to rise, with increasing investment directed toward large-scale hyperscale campuses and higher-density facilities designed to accommodate AI-driven workloads. The scale, speed and computational intensity of AI model training have substantially elevated capital requirements, while simultaneously placing mounting strain on power supply and grid infrastructure. Concurrently, private equity involvement is growing, with institutional investors deploying capital into both greenfield developments and strategic acquisitions in pursuit of long-term exposure to digital infrastructure assets characterised by stable and durable cash flows. Environmental, Social and Governance ("ESG") considerations have likewise moved to the forefront of financing and development strategies. Rising energy consumption and heightened regulatory and societal scrutiny are prompting greater investment in advanced cooling systems, renewable energy sourcing and continuous efficiency enhancements. Taken together, these developments underscore the structural shift in how data centres are perceived and financed – from traditional real estate assets to complex, capital-intensive digital infrastructure platforms that demand increasingly sophisticated, flexible and diversified funding solutions.

MOVING BEYOND TRADITIONAL REAL ESTATE FINANCING

The recharacterisation of data centres as critical digital infrastructure rather than passive real estate assets has fundamentally reshaped the way capital is structured and deployed. Unlike traditional real estate projects, data centres are characterised by exceptionally high upfront capital expenditure, often running into the billions for a single hyperscale facility. As a result, financing structures have evolved to reflect the scale and technical complexity of these assets. Data centre financings can be structured along a spectrum of established financing models, including real estate, corporate, leveraged and project or infrastructure finance principles. The approach adopted may depend on several factors, including the internal structuring expertise of the originating financial institution, the composition of the lending syndicate in club or syndicated transactions, the development stage and type of facility, and the profile and concentration of its customer base. In practice, many data centre financings are hybrid in nature, combining elements from different segments of the debt markets to accommodate both real estate characteristics and infrastructure-style risk allocation. Within these structures, senior debt provided by commercial banks and institutional credit providers typically forms the core of the capital structure given relative cost efficiency compared to other funding sources. However, the magnitude of capital required means that projects seldom rely on a single layer of financing. Instead, multi-tranche structures are common, layering senior facilities with mezzanine debt to bridge funding gaps and optimise capital efficiency. While senior debt remains comparatively economical, it is typically accompanied by robust covenant packages, which merit close attention from investors, given the operational and tenant concentration risks inherent in this asset class. The amount of capital required for data centre development is also expanding the range and profile of capital providers active in the sector. While a number of operators have been taken private in recent years, it is increasingly apparent that public capital markets will need to assume a more prominent role in financing future growth. At the same time, participation from sovereign wealth funds, pension funds and dedicated infrastructure investors is expected to intensify, reflecting both the growing funding requirements and the need to distribute risk across a broader investor base. As the financing landscape evolves, companies seeking capital will need to take a strategic and forward-looking approach to funding and investment structures to optimise their positioning.

SINGAPORE DEVELOPMENTS

Regulatory developments In Singapore, data centre operators providing digital infrastructure services foundational to the economy will be regulated as Foundational Digital Infrastructure. This imposes requirements to adhere to cybersecurity codes and standards of practice. Further, data centres designated as Critical Information Infrastructure under the Cybersecurity Act 2018 must safeguard their systems against cyberthreats by complying with codes and directions. In line with the Advisory Guidelines issued by the Infocomm Media Development Authority ("IMDA"), data centre operators are expected to put in place robust continuity management systems to safeguard against cybersecurity risk and are expected to implement business continuity policies, alongside controls and processes, in order to minimise service disruption to data centres. While these developments do not directly restrict the ability of lenders to take and enforce security, due diligence requirements will inevitably heighten, given the increasingly stringent compliance standards. Rise of portfolio financing In Singapore and the wider Asia-Pacific region, there has been a clear trend where developers are increasingly moving towards portfolio-level financing involving multiple assets. As compared to individual financing, portfolio-level financing can offer a more scalable funding model with greater structural flexibility. For example, the Keppel Data Centre Fund III has achieved a US$580 million first close to invest in a multi-asset portfolio of AI-ready hyperscale data centres all across Asia-Pacific. Advent of hyperscale data centres Financing of hyperscale data centres has been accelerated alongside the AI revolution which demands high performance computing supported by large parallel processing, vast storage, significantly higher power consumption and water infrastructure. As a result, the sector is gravitating towards hyperscalers who can provide unmatched efficiency, scalability and cost effectiveness for thousands of servers. Inevitably, smaller data centres, particularly those catering to enterprise tenants, may face diminishing competitive advantage in an environment prioritising scale-driven requirements. According to UOB Kay Hian Research’s latest report "Data Centre REITs – Size Matters!", data centres in Singapore are particularly well-positioned to support mission-critical and low latency applications, such as financial services and AI inference applications. Singapore's role as a regional connectivity hub stems from its 26 international subsea cables and three cable landing sites, linking Southeast Asia to the global network. The country is targeting at least US$10 billion in investment over the next decade to expand its subsea cable capacity and landing sites in order to support the usage of emerging AI applications. Singapore data centre financing deals In the past year, Singapore has seen an increasing number of data centre financings. Investment giant KKR has announced an agreement to fully acquire ST Telemedia Global Data Centres ("STT GDC"), a leading data centre provider with more than 95 data centres across 11 geographies and points of presence in over 20 major business markets. KKR, together with Singtel, signed definitive agreements to acquire the 82% stake in STT GDC from founding shareholder ST Telemedia for S$6.6 billion, building upon the Consortium's earlier S$1.75 billion investment in 2024. Asia-Pacific & Japan hyperscale data centre specialist, AirTrunk, also secured a landmark S$2.25 billion green loan in Singapore to support the development of a new hyperscale data centre, AirTrunk SGP2. As Singapore’s landmark green loan for a data centre, it represented a significant milestone in sustainable digital infrastructure investment regionally and further reflects the accelerating momentum toward responsible capital deployment and reinforcement of Singapore's positioning as a leading green finance hub. Other significant deals include Nxera DC, Singtel’s regional data centre arm, securing a S$643 million five-year green loan to finance the development of a 58MW data centre in DC Tuas, with the financing backing from DBS, OCBC, Standard Chartered, HSBC and UOB as green loan coordinators. Keppel also secured capital commitments of almost S$2 billion for its three flagship funds, one of which would invest in a portfolio of Asia-Pacific sustainable data centres, while mitigating investor portfolio leasing risks by securing pre-commitment or high leasing certainty from hyperscale customers. Beyond domestic financings, Singapore banks have also been involved in financing the growth of data centres in the region. For example, DBS and UOB have jointly provided a S$530 million loan facility to finance the development of a new data centre campus in Batam, Indonesia, acting as a "digital bridge” between the Singaporean and Indonesian economies.

RISKS IN DATA CENTRE FINANCING

Despite the rapid expansion of the data centre financing industry, issues remain, which companies should consider when entering or developing within this market. Credit risks Credit risk considerations are particularly important for assets financed under project finance structures, with investors typically assessing cost matrices such as land cost, Internet connectivity, power, tax environment, natural disasters, and the workforce. Beyond traditional credit risks, lenders also face sector concentration risk when financing multiple large data centres or AI infrastructure projects. Analysts have cautioned that the rapid pace of capital deployment in this space could resemble prior market bubbles, with some comparing it to the pre-dot-com credit build up. Such dynamics raise questions about the sustainability of lending trends in a sector that is both highly capital-intensive and technologically fast-evolving. Specifically, the interconnectedness between AI firms, data centre financing, and broader credit markets may magnify potential losses if valuations were to correct, increasing the systemic impact of concentrated exposures across lenders' portfolios. Reliance on cross-collateralisation Many data centre financings now rely on cross-collateralisation, where cash flows from stabilised assets (typically comprising operational data centres with long-term tenant leases and predictable revenue streams) are used to support financing for new developments. In portfolio financing structures, which combine operational and development-stage assets under a single arrangement, lenders assess the aggregate cash flows of the portfolio, using the reliable income from stablished assets to improve bankability and optimise leverage. While cross-collateralisation strengthens financing, it also introduces concentration risk – a decline in income from stabilised assets, whether due to service level agreement penalties or tenant exit, may undermine the overall financing structure and expose lenders to broader credit stress. Risks arising from lease Once a data centre becomes operational, construction risk falls away and the focus shifts to revenue generation. At this stage, creditworthiness depends on the asset's ability to produce stable and predictable cash flow sufficient to service debt. Lease structure and tenant quality therefore become central to lenders' analysis. A long-term lease to an investment-grade hyperscaler such as Amazon, Google or Microsoft can materially strengthen the asset's risk profile. With contractual income extending 15-20 years, no break options and fixed or inflation-linked rent reviews, the asset begins to resemble a long-dated corporate bond supported by strong covenant strength. From a lender’s perspective, rental income from such tenants provides visibility of cash flow over the tenor of the financing. Although data centres have exhibited relatively low tenant turnover with European markets recording a marked rise in net absorption in early 2024, the continued delivery of new capacity and intensifying competition among operators may gradually shift bargaining power toward tenants. This effect is likely to be more pronounced for older or higher-cost facilities that lack clear competitive differentiation. Despite favourable market dynamics, lenders differentiate sharply based on tenant composition and lease profiles. Assets without anchor tenants may be viewed as higher risk, particularly co-location or speculative facilities where leases are typically shorter and income is more fragmented. In such cases, lenders commonly require higher equity contributions, increased margins and additional structural protections, including debt service reserve accounts covering six to twelve months of scheduled debt service to mitigate the perceived rental profile gap. Transactions with higher revenue concentration are typically underpinned by tenants of strong credit quality, which lowers default risk and helps offset the exposure inherent in relying on a limited number of income sources. Beyond tenant credit and lease tenor, lenders also focus on preserving the enforceability and continuity of lease income throughout the life of the financing. To protect this revenue stream, it is fairly usual for lenders to require tenants to enter into Subordination, Non-Disturbance and Attornment Agreements ("SNDA"). An SNDA creates a direct relationship between tenant and lender, requiring the tenant to provide notice and an opportunity to cure landlord defaults before terminating the lease, while assuring the tenant of continued occupation following enforcement. The objective is to preserve the lease – the primary source of repayment. Often seen in large real estate financings, data centre SNDAs often reflect the asset's technical complexity. They may require any successor owner to be a "Qualified Operator" and frequently involve negotiated cure and tolling provisions where landlord defaults concern technical service obligations rather than purely monetary breaches. While lenders generally insist on SNDAs where lease income underpins credit analysis, the scope of protection is often negotiated with hyperscale tenants. Where protections are limited, lenders may rebalance risk through higher equity, increased reserves or other structural safeguards. Concentration and utilisation risk Despite the inherent risks associated with relying on a limited number of tenants, wholesale data centres continue to dominate the European market, and similar patterns are evident in Singapore. Demand for Singaporean data centre facilities is expected to mirror global trends, with large hyperscale providers potentially occupying around 40% of co-location space, underscoring the market's concentration among a few major tenants. Singapore-listed data centre REITs further highlight this trend, with hyperscalers representing approximately 67% and 75% of rental income for Keppel DC REIT and Digital Core REIT, respectively. Strategies considered when dealing with risks Data centre financing structures reflect the varying risk appetites across the project lifecycle. Some lenders prefer short-term construction financing to limit exposure to tenant procurement risks and technological obsolescence, while others target stablished assets, where long-term cash flows from executed leases underpin debt service. A particular concern is the rapid pace of AI-driven technological change, which can render older facilities obsolete or less competitive before the end of the financing tenors or lease expiry of such older data centres. Buildings designed for earlier generations of equipment may struggle to support future high-density compute workloads without costly upgrades, creating potential exposure for lenders, especially in portfolios reliant on a small number of high-credit tenants. Analysts note that critical infrastructure such as servers and networking typically refresh every three to seven years – often shorter than the legal term of data centre loans – making obsolescence a material long-term risk. Lease provisions, including value-guarantee mechanisms, help mitigate obsolescence risk by shifting upgrade and replacement obligations to tenants. By contractually allocating responsibility for technological refreshes, these clauses preserve the stability of cash flows from key tenants.

EVOLUTION IN DATA CENTRE FINANCING MODELS

The rapid expansion of the sector is driving structural shifts in how hyperscalers finance their operations. Historically, companies such as Microsoft, Amazon and Alphabet primarily relied on internally generated cash flows to fund data centre growth. However, the capital-intensive nature of AI infrastructure is straining this model. Sustained outlays necessitate diversification of funding sources, prompting greater reliance on external capital markets. For example, DayOne Data Centers Singapore Pte Ltd reportedly sought at least US$1 billion in private credit financing, reflecting growing investor appetite for opportunities linked to the AI boom. Greater institutional and cross-border capital participation There has been growing interest in Asian data centre assets from institutional investors, including global private equity, infrastructure funds, and sovereign wealth funds. Increasing capital deployment can be observed due to the sector’s durable operational cash flows and its systemic importance to digital infrastructure. This has resulted in international lenders and export credit agencies being increasingly active. These entities often partner with local and regional banks, providing large-scale, multi-currency financing packages. On the borrower side, large technology companies remain among the most active users of debt financing. The scale and acceleration of such borrowing reflect a broader structural shift – even highly profitable technology companies are increasingly turning to external capital markets to fund AI-driven infrastructure, reflecting both the capital intensity of next-generation data centres and the growing role of institutional lenders in meeting that demand, Syndicated loans as a convergence point for bank and non-bank capital Data centre financings are also increasingly diversified, extending beyond traditional bank loans where capital requirements now exceed what conventional banking syndicates can efficiently provide. While these traditional banks do retain significant market share in data centre financing, particularly those with established infrastructure and real estate practices, the funding gap has created unprecedented opportunities for private credit and non-bank lenders to participate on a large scale. Based on UBS' estimate, private credit financing for AI-linked data centres nearly doubled in the 12 months through early 2025, reflecting a marked acceleration in capital deployment within the sector. The appeal of data centre infrastructure to private credit lenders lies in the convergence of long-term contracted revenues, investment-grade tenant creditworthiness, and infrastructure-like risk profiles. On a large scale, Meta's record-breaking US$27 billion private financing joint venture with Blue Owl Capital to fund its "Hyperion" data centre campus highlights the command that infrastructure funds and mega-private-credit players now have in capital deployment capacity, rivalling traditional bank syndicates and public bond markets. Shift from corporate funding to project finance structures In addition, the industry is experiencing a shift from corporate balance sheet funding to project financing structures. While corporate funding remains common, it is likely to become less efficient for contemporary data centre projects which scale and complexity favour project finance structures. Large-scale greenfield developments demand substantial upfront capital commitments, but expose sponsors to concentrated risks in construction, power procurement, and operations. Project finance addresses these challenges by ring-fencing risks at the asset level, enabling lenders to underwrite against contracted cash flows rather than corporate credit, while simultaneously preserving sponsor balance sheet capacity for additional deployments. Increasing Graphics Processing Unit ("GPU") financing Ancillary to the shift to project finance structures are emerging classes of assets that can be utilised as collaterals in data centre financing. One such class of asset is the GPUs, an essential component that forms the bedrock of modern AI performance. However, GPUs are capital-intensive, scarce and subject to rapid technological obsolescence, which results in unique financing challenges especially for traditional financing structures. A project finance structure thus helps overcome these challenges by adopting an asset‑centric approach, enabling financiers to secure loans directly against the GPUs themselves. Aside from GPU-backed financing, other off-balance-sheet funding using GPUs also include GPU leasing or subscription models, such as Elon Musk's xAI. There, a special purpose vehicle was created for the purposes of purchasing GPUs and leasing them back to xAI. This structure allowed xAI access to USD$20 billion worth of Nvidia GPUs without direct incurring any balance sheet funding, while the majority of the risks sit with the investment firm Valor Equity Partners.

DATA CENTRE FINANCING AND ESG TRENDS

General ESG considerations The rapid expansion of AI has driven significantly greater power demand. Today, the technology sector is estimated to account for 1.5-4% of global greenhouse gas emissions, a figure projected to rise as AI adoption accelerates. Data centres, which form the backbone of the digital ecosystem, are inherently energy-intensive operations that consume substantial energy for processing and depend heavily on water for cooling. The additional strain created by AI workloads has amplified sustainability concerns, particularly regarding electricity consumption, water usage and overall carbon footprint. Opposition to projects such as Beale Infrastructure's Project Blue in Tucson, Arizona, as well as Malaysia and Taiwan's refusal of unsustainable data centre projects demonstrate how ESG considerations now directly influence project viability. For Singapore, greening information and communications technology is critical in a digital and carbon-constrained world. The digital economy contributes approximately 20% of Singapore's Gross Domestic Product, and as Southeast Asia's premier data centre hub, the country hosts critical cloud and enterprise infrastructure. Singapore has committed to achieving net zero emissions by 2050, creating a delicate balance between economic growth and climate responsibility. In response, Singapore has adopted a comprehensive and structured policy framework. Working closely with data centre operators, the government pioneered the world's first Tropical Data Centre standard – a set of guidelines designed to enable data centres to operate efficiently at higher ambient temperatures while optimising energy performance. Building on this foundation, the IMDA launched the Green Data Centre Roadmap, establishing clear targets for energy optimisation, renewable energy integration and sustainable capacity management. Under this framework, the government will allocate new capacity to operators that prioritise sustainability and economic value, while offering support schemes such as the Energy Efficiency Grant to facilitate the adoption of more efficient technologies. Through these measures, Singapore seeks not only to meet its climate commitment while sustaining digital growth, but also to catalyse innovative solutions to green data centres across the region. The Monetary Authority of Singapore ("MAS") has provided guidance for green and sustainability financing by providing guidelines on the application of the Singapore-Asia Taxonomy in the financial and corporate sectors. The Taxonomy has been integrated by local banks into their financing frameworks and will be used as an "assessment tool alongside other taxonomies or market standards when evaluating green and transition financing solutions". ESG-Linked data centre financing considerations In parallel with regulatory and community pressure, financial markets are also adjusting to the sustainability imperatives of data centre development. Lenders and institutional investors active in Asia’s data centre sector are increasingly aligning data centre financing with clear ESG-driven Key Performance Indicators ("KPIs") and Sustainability Performance Targets ("SPTs"). There has been a marked rise in the use of green and sustainability-linked financing framework where interest margins are tied to measurable ESG outcomes such as energy efficiency, carbon-free power commitments or other sustainability targets. In Singapore, the energy efficiency of data centre is measured using the Power Usage Effectiveness ("PUE") ratio under the Building and Construction Authority-IMDA Green Mark scheme. For new data centres to qualify for the "Platinum" rating, it must operate at a PUE of 1.35. These operational metrics are important as they are relied upon as part of lenders’ due diligence. Practical considerations for borrowers There are also practical issues that borrowers should be mindful of – issues that affect bankability, pricing and execution certainty. Firstly, KPIs and SPTs selection is a fundamental aspect of green financing for data centres and these criteria form part of the pricing mechanism, pursuant to which margins may step up or down. The KPIs should align with the borrower's operational strategy and be measurable or quantifiable on a consistent methodological basis, and the pricing incentives must be accompanied by performance and carry potential reputational exposure. Secondly, to ensure that the achievement of the KPIs and SPTs can be verified, borrowers must prepare and maintain market reports regarding their performance, bearing in mind that there is currently no standard market report templates, although there have been emerging methodologies of reporting such as the Global Reporting Initiative, Sustainability Accounting Standards Board and International Sustainability Standards Board. Thirdly, borrowers should stay abreast of evolving regulatory expectations. The regulatory landscape remains dynamic and authorities such as the MAS are consistently advancing systems to improve ESG data flows and disclosure expectations.

CONCLUSION

Looking ahead to the remainder of 2026, the market for data centres is expected to remain high in demand. The resurgence in capital commitment has reinforced market conviction that demand – rather than speculative exuberance – underpins sectoral growth. Strong governmental support for data centre initiatives in 2025, coupled with several private sector deals, have set the foundation for Singapore to continue to be a data centre investment regional hub in 2026. At the same time, institutions have to remain cognisant of the risks and challenges that data centre financing faces, especially so given the importance of mitigating concentration risk and the increased focus on meeting ESG goals. Nevertheless, as observed by the APLMA White Paper on Data Centre Financing published on 11 March 2026, the outlook for data centre financing for 2026 and beyond remains robust and compelling, underpinned by accelerating digital transformation and AI-driven demand.

Fintech and financial services regulatory: local firms

FOCUS ON…

Rankings

Tier 1

Allen & Gledhill LLP

Baker McKenzie Wong & Leow

Rajah & Tann Asia

WongPartnership LLP

Tier 2

Dentons Rodyk

OC Queen Street

Tier 3

Clifford Chance

Drew & Napier LLC

Shook Lin & Bok LLP

Withers KhattarWong LLP

Blackletter LLC

Prolegis LLC

Tier 0

CMS