Big Tech holds nearly $2.4 trillion in spending commitments for AI

The four largest US technology companies have locked in nearly two and a half trillion dollars in purchase commitments, contractual obligations, and leases tied to artificial intelligence, Bloomberg reported on Thursday. Alphabet leads the group with about $900 billion in total commitments, a figure roughly nine times higher than a year earlier. Meta follows with about $700 billion in future spending, Amazon has budgeted $220 billion in capital expenditure for this year alone, and Microsoft’s own obligations round out the group.

The numbers capture more than annual capex. They include long-term purchase orders for chips, power contracts, and data centre leases that stretch decades into the future, commitments that will eventually land on balance sheets but for now sit in financial footnotes. About half of Meta’s total is tied to data centre leases running as long as 30 years, roughly eight times what the company carried a year ago.

Alphabet’s commitments have grown at a similar pace, much of it structured through off-balance-sheet vehicles that keep the headline debt figures looking tidy.

Two of the four are already feeling the strain. Alphabet reported negative free cash flow of nearly six billion dollars in the second quarter, its first quarterly outflow in close to two decades. Amazon’s free cash flow swung to negative territory on a trailing-twelve-month basis earlier this year, and its second-quarter results showed the cash position remains under pressure even as AWS revenue grew 37 percent.

Meta is expected to follow. The company narrowed its 2026 capex forecast to between $130 billion and $145 billion by raising the low end, and its free cash flow fell 91 percent year on year in the second quarter. The pattern across all four companies is the same: revenue is growing, but capital spending is growing faster, and the gap shows up first in cash flow.

Amazon CEO Andy Jassy compared the current moment to the first AWS build-out, when the company spent years investing before demand caught up. The analogy is meant to reassure investors that the spending will eventually pay for itself, much as AWS grew from a side project into a business generating more than $40 billion a quarter. The difference is scale: the original AWS build-out cost a fraction of what the industry is now committing in a single year.

The question hanging over earnings season is whether demand is keeping pace. AWS answered in the affirmative this week with its fastest growth in four years, and Alphabet’s cloud unit grew 82 percent in the second quarter. But combined capex across the largest cloud operators is on track to overtake the cash their core businesses generate, a threshold the industry has never crossed before.

For now, all four companies remain profitable, because capital spending is depreciated over years rather than booked as an immediate expense. The accounting smooths the cost, but it does not eliminate it.

The $2.4 trillion in commitments is real money that will come due whether AI revenue materialises at the scale these companies are betting on or not.