Investment in global data centre infrastructure is poised to surge, with forecasts suggesting expenditures could reach between $5 trillion and $7 trillion from 2026 to 2030. JP Morgan estimates a base-case annual capital expenditure of $1 trillion, while McKinsey asserts that these figures could be even higher. This scale is significant when viewed against the backdrop of Bank of America’s assessment, which suggests that building one gigawatt (GW) of data centre capacity costs approximately $50 billion. At that rate, an annual investment of $1 trillion would enable the construction of around 20 GW of new capacity—three times New York’s installed electricity capacity of about 6.7 GW.
In Silicon Valley, hyperscalers and AI-focused companies are rapidly advancing their investments. Notably, in 2025, a network of circular financing deals emerged, with companies like OpenAI, Oracle, and SoftBank committing $500 billion to their Stargate project over the next four years. CoreWeave also signed a $14 billion deal with Meta for computing power, indicating a trend toward inter-company funding. Goldman Sachs projects that capital expenditures for AI hyperscalers might reach $394 million by the end of 2025, reflecting a robust appetite for investment in data infrastructure.
However, the financing landscape remains complex, particularly for standalone AI developers. Despite OpenAI generating approximately $20 billion in annual recurring revenue in 2025, its commitment to invest $1.4 trillion over the next eight years raises questions about funding viability. Historically, hyperscalers financed AI investments through their cash flows, and in 2025, companies like Meta, Microsoft, Amazon, and Alphabet collectively had around $500 billion in free cash flow. Additionally, their reliance on the debt capital market has increased as the technology sector braces for sustained growth.
A pertinent case study is Meta’s Hyperion Data Center in Louisiana, where a partnership with Blue Owl Capital was established to develop the project through a special-purpose vehicle (SPV) known as Beignet Investor. According to reports, this SPV has successfully raised $30 billion, consisting of $27 billion in loans from private credit funds and $3 billion in equity from Blue Owl.
Looking ahead, Morgan Stanley projects that capital expenditures in data centres could reach around $3 trillion by 2028, with half potentially covered by hyperscalers’ cash flows. The bank estimates an additional $200 billion could be financed through corporate debt issuance, specifically through data centre asset-backed securities (ABS) and commercial mortgage-backed securities (CMBS).
Data centre financing has not been absent from the securitisation market but remains underutilised. The first data centre ABS was issued by Vantage Data Centers in February 2018, raising $1.125 billion to support expansion in key US markets. In 2021, Blackstone issued the first CMBS to finance its $10 billion acquisition of QTS Realty Trust.
For computing power ABS and potentially more exotic financing vehicles to emerge at scale, clearer evidence of AI monetisation might be required.
In Europe, the securitisation landscape is still in its infancy, with only two ABS deals completed so far. Vantage issued its first European data centre ABS in June 2024, raising £600 million for two data centres in Wales. A year later, another €640 million was raised for four data centres located in Germany, marking a significant step for the market.
In 2025, 27 data centre ABS deals were issued, collectively raising $13.3 billion and marking a 55 percent increase year-over-year. Currently, data centre securitisation is primarily issued by operators, relying on long-term cash flows from lease payments. The proceeds are often reinvested to refinance existing debt and expand capacity.
As financing methods evolve, neocloud providers are emerging, allowing AI developers to rent computing power rather than incur substantial upfront capital expenditures. The GPU-as-a-Service (GPUaaS) model transforms infrastructure spending into flexible operating costs, paving the way for new financing opportunities. For example, Amsterdam-based Nebius has secured long-term agreements with Microsoft and Meta, demonstrating the potential for predictable cash flows that could facilitate securitisation.
Despite these advancements, recent months have shown signs of strain within the AI investment landscape. Oracle’s stock plummeted by 43 percent by late December 2025, reflecting concerns about the company’s ability to meet its ambitious AI infrastructure commitments. Its 10-Q report revealed $248 billion in lease commitments, raising alarms about the mismatch between the duration of these obligations and contracted revenues.
The web of circular financing in 2025 has brought some of the neocloud providers to the fore.
Given these dynamics, market sentiment remains cautious. Nonetheless, experts like Sung Cho from Goldman Sachs express optimism about AI funding, noting that 90 percent of the capital expenditure to date has been sourced from hyperscalers’ operating cash flows. Achieving a viable B2B2C model, demonstrating the monetisation potential of AI, may provide the foundation for mainstream adoption of computing power ABS deals, diversifying financing options in the sector.
See also
Finance Ministry Alerts Public to Fake AI Video Featuring Adviser Salehuddin Ahmed
Bajaj Finance Launches 200K AI-Generated Ads with Bollywood Celebrities’ Digital Rights
Traders Seek Credit Protection as Oracle’s Bond Derivatives Costs Double Since September
BiyaPay Reveals Strategic Upgrade to Enhance Digital Finance Platform for Global Users
MVGX Tech Launches AI-Powered Green Supply Chain Finance System at SFF 2025
