SemiAnalysis, a semiconductor and AI infrastructure research firm, has projected that AI compute spending will reach a cumulative $11 trillion through 2029, with $7 trillion of that financed by debt.
July 6, 2026 · SemiAnalysis Forecast
The $11 Trillion AI Compute Bill — And Who's Financing It
AI compute spending is projected to reach roughly $11.1 trillion cumulatively from 2024–2029 — with more than $7 trillion financed through debt, poised to become the second-largest asset-backed debt market after U.S. mortgages.
$11.1T
Cumulative AI Capex, 2024–2029
$7.1T+
AI debt balance through 2029
$2T+
Annual Capex by 2028
A New Debt Mountain
The projected $7.1T AI debt balance vs. the largest asset-backed debt market — U.S. mortgage-backed securities (~$13T). Column height is proportional.
$7.1T+
AI-related debt (2029)
~$13T
U.S. mortgage-backed securities
The "AI Project Trinity"
The three essentials to build an AI cluster — with Nvidia's revenue-guarantee backstop acting like a central bank to draw in lenders.
Capital
Debt financing at scale
→
Offtake
Long-term usage contracts
→
Datacenter
Securing the facilities
How Forecasts Compare
SemiAnalysis cumulative Capex · 2024–2029
~$11.1T
SemiAnalysis AI debt balance · through 2029
$7.1T+
McKinsey AI datacenter investment · through 2030
$5.2T
ARK Invest annual investment · through 2030
$1.5T
The Bull Case
Nvidia's backstop opens the market amid strong short-term rental demand — and build "speed" becomes a competitive moat.
The Risks
Reliance on hyperscaler backstops, banks' weak grasp of AI TCO, no benchmark for GPU residual value — and ~$7T of concentrated debt.
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article…