Markets MSFT +2.78% JUL 15, 2026

Big Tech's $700B AI Capex Print Is Landing — SMBs Still Stuck in Pilot Purgatory

Microsoft's AI run rate hit $37B up 123%, Google Cloud grew 63%, and hyperscaler 2026 capex is tracking near $700B. IDC says SMBs are still hunting pragmatic, ROI-first tools.

BY HENLEY MARRAST · MARKETS DESK

Microsoft’s March-quarter 8-K put $82.9 billion in revenue on the tape, up 18% year-over-year, with $38.4 billion in operating income and $4.27 in diluted EPS. Buried inside the print: Satya Nadella confirming the AI business “surpassed an annual revenue run rate of $37 billion, up 123% year-over-year.” Microsoft Cloud alone did $50 billion in the quarter. The capex, in other words, is converting.

Google Cloud reinforced the story a few days later, growing 63% to $20 billion, per Fortune, with Alphabet’s enterprise cloud backlog swelling to $462 billion, nearly double the prior quarter. Alphabet CFO Anat Ashkenazi described “unprecedented internal and external demand for AI compute resources.” That’s the phrasing of a CFO who wants Wall Street to keep funding the buildout.

And the buildout is enormous. CNBC pegs combined 2026 hyperscaler capex near $700 billion, more than 60% above 2025. Microsoft has guided $190 billion for the calendar year, of which CFO Amy Hood attributes roughly $25 billion to rising memory and component costs. Alphabet raised its range to $180–$190 billion, Meta lifted to $125–$145 billion, and Amazon is expected to spend $200 billion, with a recent SEC filing signaling it may tap both equity and debt to fund it. Barclays already models negative free cash flow at Meta in 2027 and 2028.

Goldman Sachs now tracks roughly $7.6 trillion in cumulative AI capex from 2026 through 2031, and lifted its estimate for the four largest hyperscalers through 2030 from $4.5 trillion to $5.3 trillion after the Q1 prints. The scale is starting to look less like a cycle and more like a fixed feature of the S&P’s capital structure, echoing the postwar telecom buildouts that took a decade to earn their return.

The catch sits downstream. IDC’s Worldwide Small and Medium-Sized Business 2026 Predictions describes SMBs moving from experimentation to adoption, but demanding “pragmatic use cases that are easy to deploy and deliver measurable ROI,” with FinOps discipline now essential. That’s a market the hyperscalers aren’t really serving. It’s the opening being worked by no-code, model-agnostic entrants like LemonLime’s “company brain” platform, which package the capex-funded models into something a 40-person business can actually deploy on Monday.

Seven hundred billion buys a lot of GPUs. It doesn’t, on its own, buy adoption.

Sources

About the author

Henley Marrast

MARKETS DESK

Henley Marrast covers AI-equity flow, accelerator demand, and earnings prints for AI Sheet Report. She leads coverage of the public AI complex from the New York markets desk, with a focus on the daily tape and quarterly results. She has been writing about technology markets for several years.