Infocomm Media Development Authority (IMDA) launched the Singapore Green Data Centre Roadmap 2024 in June 2024, which is a national strategy to ensure sustainable growth of data centres while supporting Singapore’s ambitions as a digital and AI hub. The roadmap looks to address the high energy and resource intensity of data centres, to enable continued capacity expansion without compromising sustainability and to support Singapore’s Green Plan 2030 and net-zero emissions goal by 2050.
A core target of the Singapore Green Data Centre Roadmap 2024 is to add at least 300MW of new data centre capacity in the near term, with more possible through green adoption.
DC-CFA2 criteria
To progress the Singapore Green Data Centre Roadmap 2024, IMDA together with the Singapore Economic Development Board (EDB) launched DC-CFA2 (Data Centre – Call for Application) on 1 December 2025. This initiative would look to allocate at least 200MW of new data centre capacity, with potential for more depending on implementation for innovative green energy solutions.
The application window for DC-CFA2 closed on 31 March 2026. The market is now awaiting announcements regarding successful applicants. With the DC-CFA2 application window now closed, attention has shifted from eligibility requirements to the practical challenges of delivering compliant projects.
The DC-CFA2 application followed the allocation of 20 hectares of land by Singapore on Jurong Island for the development of a low-carbon data centre park which can potentially support up to 700 MW of data centre capacity. Whilst DC-CFA2 did not mandate Jurong Island as the only location for construction of data centres, it would be a prime site for qualifying projects. It is Singapore’s energy and chemicals hub hosting infrastructure for natural gas, biofuels, and potential hydrogen/ammonia supply chains, which would be it easier for data centres to tap into the green energy pathways required under DC-CFA2 and the Singapore Green Data Centre Roadmap 2024. More generally, Jurong Island offers dedicated land parcels for industrial-scale projects which would allow for hyperscale deployment.
DC-CFA2 sets out three broad qualification criteria: strategic, economic and sustainability1:
| Strategic Criteria | Applicants should demonstrate how the proposed data centre would enhance Singapore’s position as a digital and AI infrastructure hub and improve international connectivity and digital infrastructure resilience. |
| Economic Criteria | Applicants should indicate the quantum of fixed asset investments and commit to activities that drive innovation, R&D, product development, and talent development. The application should demonstrate how the project will improve competitiveness of industries and companies based in Singapore. |
| Sustainability Criteria |
Applicants should demonstrate how they would run the most efficient data centre that is best in class:
The applicants are also required to maximise the use of green energy to power their proposed new data centre capacity, with at least 50% powered by eligible green energy pathways such as biomethane, low-carbon ammonia, low-carbon hydrogen, novel fuel cells with carbon capture & storage technology, or vertical building-integrated photovoltaics/building-applied photovoltaics. |
Key challenges
Below are some key challenges that applicants under the DC-CFA2 program are expected to have to address:
- Ultra-Low PUE at Full IT Load – Meeting the PUE of ≤ 1.25 at 100% IT load would require advanced cooling and highly efficient electrical systems. Singapore’s hot and humid climate would make cooling energy-intensive.
- Securing green energy pathways – The DC-CFA2 criteria requires at least 50% of the proposed new data centre capacity to be powered through eligible green energy pathways (biomethane, hydrogen, ammonia, solar). Given Singapore’s limited renewable generation and land constraints, the future scalability of Singapore's data centre sector increasingly depends on regional renewable electricity imports and ASEAN grid interconnection initiatives. It is also useful to note that Singapore is running a biomethane sandbox to test commercial viability and operational frameworks for imports.
- Land constraints – Whilst Jurong Island offers advantages as noted above, space is limited and competition for plots will be intense. Building data centres outside Jurong would mean that the operators would need to independently secure green energy and cooling solutions.
- Designing for AI ready high density computing – Hyperscale cloud providers need large scale AI ready infrastructure. DC-CFA2 sites may offer high-density racks (≥100 kW per rack), which would need specialized cooling systems, structural changes and new operational expertise, all of which would impact the overall project costs and operating expenses of the data centre.
- Water Usage Effectiveness (WUE) – This metric measures the amount of water used per MW of IT energy consumed. The industry benchmark WUE for data centres is approximately ≤ 2.0 m³/MWh. To meet these requirements would involve implementing advanced water recycling and alternative cooling fluids which would add to operating costs of the data centres.
- Digital twin integration and smart grids – While not a strict requirement under DC-CFA2, integration of a digital twin for predictive energy and thermal management would enable real-time monitoring and simulation which would be able to detect issues before they cause downtime and simulate load changes and cooling strategies to maintain PUE targets. Incorporating this element requires advanced sensors for real-time monitoring of power, cooling, and IT loads and IoT-enabled infrastructure for continuous data collection. The cost of hardware, software licenses and integration would add to the overall project costs. As well, developers may increasingly seek to incorporate grid-smart features, in other words, capable of dynamic load management and integration with Singapore’s energy ecosystem. Energy storage may be incorporated for demand response. This would add complexity to the electrical design of the data centre and require coordination with grid operators and regulatory compliance, increasing project timelines and costs.
Compliance with the requirements noted above involve specialized technology, partnerships, and R&D.
Financing structures for data centres
We set out potential financing structures which developers of DC-CFA2 data centres may consider. In relation to data centre financings, there is no ‘normal’ and in each case hybrid terms to suit the particular commercial drivers and sensitivities are typical. It is important therefore to be able to navigate experiences of the various different models in order to find solutions that address the interests of all stakeholders.
Real estate finance
Some lenders have traditionally viewed data centres through a real estate finance lens, with significant focus on the value of the land and physical property rather than operational performance (although ultimate structure, tenor and pricing being influenced by the degree of any pre-letting). This approach influences how risk is assessed and capital is priced, often resulting in different cost of capital expectations compared to infrastructure financiers.
This type of financing is especially suitable for acquiring land and funding early-stage development, where no customer contracts are in place (or indeed single asset development financings which are pre-let) and the asset’s value is largely tied to its physical footprint.
Project finance (non-recourse or limited recourse)
Project finance on a non-recourse or limited recourse basis aligns with infrastructure-style financing and stable post-completion revenues.
In the case of a single asset project finance, financing would typically be provided to a special purpose vehicle (SPV) on a limited recourse basis and the lenders would primarily rely on the revenues generated by the project itself. These revenues are usually underpinned by one or more long-term offtake contracts - often with a hyperscaler - which provide the necessary bankability for lenders to commit capital. When financing a portfolio of greenfield data centres, the structure may resemble single asset financing but introduces additional complexity. For instance, lenders must consider whether a default in one project should trigger a drawstop or cross-default across the entire portfolio. This depends on the relative importance of the defaulting project within the overall structure. Typically, financial covenants are tested at the portfolio level, rather than on a per-project basis, allowing for greater flexibility and risk distribution across multiple assets.
Corporate finance
This financing structure involves debt at the operating company level, whether secured or unsecured. It would be relevant where an existing data centre platform or portfolio financing is being contemplated. This type of financing would allow for flexible covenants, often backed by parent guarantees. It can be executed swiftly.
Blended financing structures
In certain cases, financing may be structured to evolve over time through a blended financing model. For example, where the data centre project is in a development phase with no signed customer contract, a real estate finance approach may be initially adopted. To ensure long-term flexibility thereafter, the financing can be designed to transition to an infrastructure-style regime once a contract with an eligible customer is secured. This shift reflects the changing risk profile and bankability of the asset - triggering changes such as the fall-away of certain sponsor guarantees upon customer contract signing.
Bridge financing
Bridge financings are used to address timing mismatches in data centre finance transactions, particularly in the form of a "bridge to institutional take-out". This structure usually involves a commercial bank or other financial institution providing a short-term loan, with the expectation that the loan will be repaid once long-term institutional capital - typically from pension funds, insurance companies or infrastructure debt funds - is deployed. Although institutional investors may have committed to the transaction, procedural delays, regulatory approvals or internal decision-making timelines may prevent immediate execution. The bridge loan serves as a temporary solution enabling sponsors to meet urgent financing needs without waiting for institutional capital to be available. These facilities are usually structured with a clearly defined exit strategy and a short maturity period - often between a few months and a year. While the cost of capital may be higher than traditional long-term debt, the speed and flexibility of bridge financing make it a valuable tool in time-sensitive situations.
Green and sustainability linked financing
Data centre projects under DC-CFA2 would be ideal candidates for green and sustainability-linked loans because they inherently meet ESG criteria, particularly given the emphasis on energy efficiency metrics (PUE and WUE) and procuring power through green energy pathways. Operators would need to develop a Green Finance Framework early, secure third-party certifications, and negotiate margin-linked KPIs with potential financiers.
Private credit
Growing role of private credit funds and institutional investors offering bespoke financing for hyperscale and AI-driven data centres. Whereas commercial bank lenders may hesitate due to complexities associated with the data centres under DC-CFA2 (given the green energy pathways and AI-ready specs requirements), private credit funds may be comfortable with bespoke structures and higher risk-adjusted returns. Private debt may be combined with green/sustainability linked features and co-investments from infrastructure equity funds and possibly export credit agency backed financing for the supply of imported kit.
AI infrastructure financing
Data centre financing typically focuses on the “shell” - the land, building structure, and essential infrastructure like cooling, power, connectivity, and security - excluding the “core” components such as servers, chips, and racks. In hyperscale data centres, it is common for customers to install and operate their own equipment. In contrast, AI-focused data centres often see developers owning both shell and core, offering compute capacity through models like GPU-as-a-Service, Compute-as-a-Service, or Infrastructure-as-a-Service. This approach significantly increases financing needs as the developer will not only need to make investments to build the shell but will also have to acquire and install the core. The value of the core is in most cases roughly four times higher than the value of the shell and therefore the AI infrastructure (or equipment) financing is usually very significant. Financing of equipment (the core) brings with it a different risk profile. In this fast-paced sector, hardware obsolescence is an evident risk. To match the lifecycle of the equipment and in view of the technology risk AI infrastructure financings tend to have a short term (typically five years at most) and will typically amortise, either fully or with a relatively small tail. The chips, servers and racks can be expected to be held as collateral by the financiers.
Equity financing
In addition to debt-led options, sponsors may choose to capitalise a project SPV with shareholder funds, aligning risk and return directly with project performance. This approach dovetails with the non-recourse / limited recourse approach commonly used where revenues are project-level and lenders look to the SPV’s cash flows and contracts for bankability. Equity can be raised on a standalone sponsor basis or through joint ventures to spread development, technology, and offtake risks. This approach may be attractive to sponsors given the broad qualification criteria which may mean sponsors / investors wish to engage in capability sharing and risk-sharing with partners, and may be attractive to foreign investors in particular insofar as partnering with a Singapore based group can help to de-risk the project by allowing foreign investors to leverage on their local partners’ regulatory and permitting know‑how and pre-existing relationships across the energy and infrastructure sectors.