The $700 Billion Question: Big Tech's AI Capex Arms Race Keeps Escalating
The numbers coming out of Big Tech's data center budgets have stopped sounding like capital expenditure lines and started sounding like national infrastructure programs. Combined 2026 spending plans across the four largest hyperscalers — Microsoft, Amazon, Alphabet and Meta — have climbed toward the $700 billion mark, according to reporting from Yahoo Finance and CNBC, a figure that would have seemed implausible even at the height of last year's AI enthusiasm.
The individual line items explain how the total got there. Amazon posted $44.2 billion in capital expenditures in the first quarter alone, the largest of any hyperscaler, as AWS growth accelerated to 28 percent — its fastest pace in 15 quarters, per Yahoo Finance. Microsoft reported $30.88 billion in fiscal third-quarter capex, up 84 percent year over year, alongside an AI business it says has reached a $37 billion annual revenue run rate, up 123 percent. Alphabet's first-quarter capex of $35.67 billion more than doubled from a year earlier, with a Google Cloud backlog the company put at more than $460 billion.
Meta has been the most aggressive in raising its own bar. The company lifted its 2026 capex guidance to a range of $125 billion to $145 billion, up from a prior $115 billion to $135 billion, citing higher component pricing and data center costs. The market's verdict was immediate: shares fell 9.25 percent in the session after the announcement, according to Yahoo Finance's tally.
That reaction captures the tension running through this earnings cycle. CNBC reported in late July that Amazon, Meta and Microsoft all faced skeptical investors after Alphabet's spending disclosures sparked a sell-off, and CNBC has separately calculated that the near-$700 billion spending wave is taking a visible bite out of Big Tech's cash generation. The starkest data point belongs to Amazon, whose trailing free cash flow has fallen roughly 95 percent to $1.2 billion — a number that would be alarming at any company not sitting on AWS.
The bulls have their own exhibits. Microsoft's commercial remaining performance obligations — contracted future revenue — have nearly doubled to $627 billion, suggesting demand is being booked ahead of capacity, not chasing it. Bernstein lifted its Microsoft target to $660 from $647 on Monday while keeping an Outperform rating, with the company guiding to roughly 45 percent constant-currency Azure growth this quarter, per ts2.tech. And Nvidia, the arms dealer to all sides of this race, reported $62.31 billion in quarterly data center revenue, up 75 percent, with networking revenue up 263 percent.
The capex race is also reshaping the market's second and third tiers. Data center landlord Equinix has attributed roughly 60 percent of its largest recent contracts to AI-driven deals, and Amazon's in-house chip business has reached a $20 billion revenue run rate with triple-digit growth — a reminder that some of this spending is aimed at reducing dependence on Nvidia rather than enriching it.
For the broader market, the stakes are hard to overstate. AI infrastructure spending has become one of the largest single swing factors in S&P 500 earnings and in the industrial economy itself, touching utilities, construction, power equipment and semiconductors. Analysts at Futurum Group, who peg the 2026 buildout near $690 billion, describe it as an infrastructure sprint with no modern precedent.
The question hanging over the fall earnings season is no longer whether the hyperscalers will spend — the guidance settles that — but how long shareholders will treat nine-figure daily outlays as investment rather than overhang. Meta's 9 percent single-day repricing suggests patience, while still ample, is no longer unconditional.
