How will the AI infrastructure boom be financed?

AI infrastructure spending is becoming a cross-asset financing cycle that will reward investors who distinguish durable cash flows from capital-intensive projects with weak economics.

• Anthropic reported $16.3 billion of revenue in the first half of 2026, nearly 14 times the prior-year figure, while spending nearly $9 billion on computing and infrastructure during the same period.
• Anthropic has outlined almost $800 billion of spending commitments over the next decade, with potential IPO valuations ranging from $1.2 trillion to $1.75 trillion.
• Hyperscaler CapEx is projected at approximately $697 billion in 2026, while annual AI infrastructure spending could rise from roughly $765 billion currently to $1.6 trillion by 2031.
• Alphabet guides to $195 billion to $205 billion of 2026 CapEx versus $91.5 billion in 2025, Amazon plans approximately $220 billion, Microsoft $175 billion and Meta $135 billion to $145 billion. Russell estimates 2026 free cash flow of negative $26 billion for Alphabet, negative $61 billion for Amazon, positive $67 billion for Microsoft and negative $8 billion for Meta.
• Bond markets may finance up to one-third of AI CapEx, private credit outstanding to AI-related companies has exceeded $200 billion, and another $800 billion of data-center financing could come from private credit over the next two years. CoreWeave's $7.5 billion GPU-backed facility and Meta's Hyperion project, supported by about $27 billion of external debt, illustrate specialized structures.

Takeaway: Investors can access the buildout through equity, public debt, private credit, infrastructure and structured finance, while portfolio construction must assess valuation, collateral, repayment capacity, refinancing risk, execution and overlapping exposure across managers.