Securitisation: a now essential cog in the digital infrastructure super-cycle
There’s no polite way to say it: the world is nowhere near ready for the scale of digital infrastructure that the AI era is going to require. At the same time, the financing mechanisms we’ve come to rely on are also evolving in parallel. If we’re truly aiming to build the backbone of an AI economy within this shifting landscape, there’s no doubt that securitisation will have to play a far bigger—and far more creative—role than it ever has before.
Morgan Stanley estimates that total AI‑driven infrastructure investment could reach as much as US$3 trillion over the next five years, as the world struggles to meet the sheer scale of AI demand. According to McKinsey, rising AI workloads are expected to increase global demand for data centres by 50% by 2030, creating an urgent need for costly upgrades to power, cooling, and network infrastructure.
Capital allocation to this asset class is accelerating in response to surging demand for compute. At the same time, the financing structures needed to deploy capital at the required scale are also continuing to develop. Hybrid models that bring together traditional and alternative financing approaches are emerging, though their application is still expanding in line with market needs. Bloomberg estimates that AI-related companies and projects raised around US$200 billion from debt markets in 2025, with the growing prevalence of private deals suggesting the true figure is likely higher. Looking ahead, Morgan Stanley expects AI hyperscalers alone to account for between US$250 billion and US$300 billion in debt issuance in 2026.
With US$870 billion in new debt financing still needed (according to JLL), project finance, structured finance, private credit and green financing all have a crucial role to play if the needs of this infrastructure super-cycle are to be met. Given the sheer scale of data centre developments, funding typically relies on a blended capital stack—combining project finance during construction, term lending post-completion, and, in many cases, capital markets issuance as a take-out—underscoring the continued importance of traditional financing approaches. Within this framework, securitisation and structured finance can complement existing models by broadening the investor base and introducing greater flexibility. Building an AI economy at speed and scale ultimately calls for a step-change in how capital is mobilised—one that a reimagined approach to securitisation can help to support.
How ABS facilitates market access
Digital infrastructure projects blend a variety of proven financing methods throughout the project lifecycle: the construction phase adopts characteristics of traditional project finance, including milestone-based funding structured with a payment-in-kind (PIK) mechanism. Post-construction, these become amortising term loans incorporating regular cash interest payments.
Securitisation is very useful in financing collateral backed by stable, predictable, cash-flows with little to no remaining construction or lease-up risk. Data centres, typically underpinned by long-term leases, provide a cash flow profile that resonates with market participants, analogous to other hard/real assets such as intermodal and aviation asset classes. Coupled with the structural protections inherent in securitisation vehicles these financing structures provide an ideal solution for refinancing these term loans at favourable spreads in the capital markets.
Many projects have made good use of the mechanism already, and volumes of asset-backed securitisation (ABS) backed by data centre assets are projected to reach US$50 bn in 2026.
Data centre-backed commercial mortgage-backed securitisation (CMBS) deals are more of a real estate play, secured on the mortgage of the facility itself. As such, CMBS is mostly used by large, single tenant hyperscale-type projects. Nevertheless, on a risk adjusted basis, both ABS and CMBS structures provide investors with structural protections and credit enhancement at an attractive tranche-specific spread compared with pure-play corporate debt exposure.
The problem of power
But while the cashflows may be attractive, to meet its extraordinary demand projections, data centre infrastructure must first address several important challenges.
a) Access to the grid
Data centres are getting larger (today, 200-megawatt facilities are increasingly common), and so is the amount of energy they consume. According to Barclays, US data centre power demand will grow by 14% to 21% every year until 2030. That is equivalent to 13% of all current US electricity demand by 2030.
Grid connection is a challenge, and four-year delays are already common in major markets, according to JLL. Original Equipment Manufacturers (OEMs) and operators are, therefore, now looking at off the meter solutions for energy procurement. This means that they will directly fund their own energy generation. This could become an increasingly important part of project viability in future.
b) Inflation and construction costs
Even before the recent Middle East oil shock, inflation had driven construction costs for these projects to new heights. According to the Turner & Townsend 2025 – 2026 Data Centre Construction Cost Index, there was a 5.5% increase in the cost per watt of building a traditional cloud-based, air-cooled data centre last year. However, 60% of survey respondents expect construction costs to increase by 5 to 15% in 2026.
c) Fibre connectivity
High quality fibreoptic connectivity is the backbone of any data centre project. Centres need resilient, diverse fibre routes to minimise the risk of redundancy and technological obsolescence. Fibre connectivity is as important as energy procurement in the viability of these projects, while ABS has also been used to help enterprise fibre providers finance build-out.
Sustainable solutions
Renewable energy sources can also help to address grid constraints for data centres. Indeed, JLL calculates that combining renewables and private wire transmission can lower power costs by 40% in Europe, the Middle East and Africa (EMEA). By 2030, combining Battery Energy Storage Systems (BESS) with solar solutions could become more competitive than fossil fuels, and nuclear solutions are also being considered.
There is potential for sustainable financing to become an important part of the data centre financing toolkit. Green and sustainability-linked bonds and loans based on key performance indicator (KPI) metrics, such as a reduction in greenhouse gases and waste, will bring another important pool of investors to the funding mix.
Private markets too have a transformational role to play. Data centre risk is a natural extension for private credit, and both private equity and debt funds are allocating significant amounts of capital to this sector. For example, a consortium of leading private firms has launched the Global AI Infrastructure Investment Partnership (now the AI Infrastructure Partnership) to raise up to US$100 billion to build AI data centres alongside renewable energy and storage infrastructure.
Oversold or underbuilt?
The demand projections for digital infrastructure are so large that questions are inevitably being raised around whether meeting them is even possible. Are the risks involved being properly addressed? Will the sources of both power and financing be sufficient? For the answer to those questions to be yes, immediate risks around energy procurement must be addressed by developing off-grid solutions and by clustering data centre development geographically close to energy sources. All large infrastructure development involves long lead times and significant construction risk, but the need for individual nations to assert digital autonomy and keep sensitive data within their own borders will help digital infrastructure projects to get over the line.
Securitisation won’t solve everything—but it solves a lot
No financing model can future‑proof a sector moving this quickly. But securitisation offers something extremely valuable: a proven, transparent, scalable framework for monetising long‑term digital‑infrastructure cash flows. And in a world where demand for AI is accelerating faster than anyone expected, scalable solutions are exactly what we need.
