AI Capex: The $700 Billion Bet and Its Flow Implications

The sheer magnitude of the AI capital expenditure sprint is staggering. The five largest US cloud and AI infrastructure providers – Microsoft, Alphabet, Amazon, Meta, and Oracle – have collectively committed to spending between $660 billion and $690 billion on capital expenditure in 2026. That figure represents a near doubling of their 2025 spending levels, marking an unprecedented acceleration in infrastructure deployment.

This massive bet is being funded increasingly through debt. Hyperscalers have already issued roughly $121 billion in new debt in 2025 alone, with over $90 billion raised in just the last quarter of that year. The trend points to a continued reliance on bond markets, with analysts projecting as much as $900 billion in new tech sector debt issuance for 2026.

Critically, this spending has consistently outpaced market expectations. Analyst consensus estimates for AI hyperscaler capex have missed actual spending by over 30 percentage points in both 2024 and 2025. This persistent underestimation underscores the difficulty in forecasting the true scale of the build-out, framing the entire $700 billion commitment as a high-stakes, debt-funded wager whose ultimate success hinges entirely on future cash flow generation.

The Cash Flow Reality Check

The scale of the infrastructure bet is so vast that the revenue it aims to monetize looks like a rounding error. The five largest US cloud providers are spending $660-$690 billion on AI infrastructure in 2026. Meanwhile, the pure-play AI vendors they are building for, like OpenAI and Anthropic, are growing rapidly but remain tiny in comparison. OpenAI's annual recurring revenue hit about $20 billion last year, and Anthropic's run rate recently surpassed $9 billion. Combined, their revenues are still a fraction of the trillion-dollar commitment.

This creates a stark sustainability question. The core warning from Bank of America is that capital spending is still greater than revenue growth. The industry is funding a massive build-out with debt, betting future returns will materialize. But the path to those returns is not yet clear, creating a classic "air pocket" risk where monetization lags far behind the capital deployed.

The debt math is the ultimate reality check. The $8 trillion of cumulative capex implies a need for roughly $800 billion in profit just to service the interest burden. That return path is entirely speculative today. For now, investors are buying the dream of future cash flows, while the current financial flows show a massive gap between spending and earnings.

Market Rotation and the Path Forward

The market is adapting to the new reality, with clear signs of investor rotation. After a broad rally in AI infrastructure stocks, performance has diverged sharply. Investors have rotated away from AI infrastructure companies where operating earnings growth is under pressure and capex spending is debt-funded. This shift reflects a growing selectivity, as the initial euphoria gives way to scrutiny of cash flow sustainability.

The next phases of the AI trade are expected to involve different beneficiaries. Goldman Sachs Research expects the next phases of the AI trade to involve AI platform stocks and productivity beneficiaries. This signals a flow allocation shift from pure infrastructure plays toward companies that can demonstrate a clearer link between AI investment and future revenue, such as software and services providers.

Yet, turbulence is in the forecast. The market is bracing for an "air pocket" as monetization remains uncertain and power becomes a critical bottleneck. Bank of America warns of an "air pocket" as monetization is to be determined. This sets up a volatile transition, where the focus moves from funding the build-out to proving it generates returns, a process that could see significant volatility in both infrastructure and platform stocks.