A $1.5 trillion financing gap looms as hyperscalers race to build AI infrastructure, with most of the shortfall expected to come from private credit markets
The price tag for the AI revolution just got a number attached to it, and it’s large enough to make central bankers squint. Morgan Stanley Research estimates that cumulative global data center capital expenditure will hit approximately $2.9 trillion between 2025 and 2028, with a massive chunk of that funded by debt rather than corporate piggy banks.
Where the money goes, and where it comes from #
Morgan Stanley’s breakdown splits the total into two major buckets. Around $1.3 trillion is earmarked for physical infrastructure and construction costs, the bricks-and-mortar side of building massive computing facilities. The remaining $1.6 trillion goes toward IT hardware, meaning the chips and servers that actually do the computing.
The hyperscale giants, Amazon, Microsoft, Google, and Meta, are expected to cover roughly $1.4 trillion of the total through their own internal free cash flows. That leaves a financing gap of approximately $1.5 trillion.
Morgan Stanley’s report, titled “Bridging a $1.5tr Data Center Financing Gap,” maps out where that money is likely to come from. The lion’s share, roughly $800 billion, is projected to flow from private credit markets. Another $200 billion is expected from corporate debt issuance, with $150 billion coming through securitized products.
The leverage question #
Morgan Stanley’s analysis flags the obvious risk: what happens if AI monetization doesn’t materialize at the pace these capital commitments assume? The report raises concerns about potential widespread financial risks if growth in AI-driven workloads underperforms expectations.
The knock-on effects extend well beyond tech balance sheets. Morgan Stanley’s assessment highlights significant implications for power generation, construction, and the broader financing ecosystem.
What this means for investors #
Bitcoin mining companies have been increasingly pivoting their existing infrastructure toward AI and high-performance computing workloads. As the demand for data center capacity intensifies, miners sitting on power purchase agreements and physical facilities could find themselves holding surprisingly valuable real estate in the AI gold rush.
Morgan Stanley’s assessment notes the $2.9 trillion capex cycle has the potential to meaningfully influence US GDP growth. When $800 billion in private credit flows toward a single thesis, the correlation risk becomes significant, with private credit funds, pension allocations, and insurance company portfolios all increasingly exposed to the same underlying assumption: that AI infrastructure will generate returns sufficient to justify this generational buildout.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our