Microsoft's 3.2% slide on Monday shows how quickly AI infrastructure questions can move a mega-cap stock.
Microsoft's 3.2% slide on Monday shows how quickly AI infrastructure questions can move a mega-cap stock.

Microsoft shares fell 3.2 percent on Monday after a Guardian investigation questioned whether the company has enough advanced AI chips installed to support its rapid data-center expansion. The report pointed to a potential gap between Microsoft's stated infrastructure buildout and the number of AI accelerators deployed across its facilities.
"They are giving insufficient context," Shaolei Ren, a professor at the University of California, Riverside, said of Microsoft's capacity disclosures, noting that the company's audited sustainability reports provide another measure of its infrastructure.
The Guardian reported Microsoft had 2.2 million AI chips installed globally, based on internal documents reviewed by the publication, compared with an earlier target of 1.8 million by end-2024. Microsoft said it added 5 gigawatts of data-center capacity over the past two years and operates hundreds of facilities across five continents, but reports capacity largely in terms of power rather than individual chips, complicating independent verification.
The selloff comes days after JPMorgan raised its December 2027 price target on Microsoft to $625 from $550, maintaining an Overweight rating. Analyst Samik Chatterjee cited accelerating Azure and Microsoft 365 Commercial Cloud growth tied to AI infrastructure investment, and estimated stronger Copilot adoption could generate $24 billion to $41 billion in additional revenue.
Microsoft rejected the Guardian's methodology, saying the estimates drew "the wrong conclusions from incorrect assumptions." The company said it does not disclose the volume of specific chips used in its AI infrastructure.
The dispute centers on how to measure AI capacity. Microsoft reports data-center expansion in gigawatts of power, not chip counts. The Guardian's figure of 2.2 million installed AI chips was derived from internal documents, but Microsoft argues this undercounts actual deployment because some chips sit in inventory awaiting facility completion.
CEO Satya Nadella previously acknowledged this constraint. "It's not a supply issue of chips," he said, noting that chips can sit unused because of a lack of completed data-center facilities and available electricity. The Guardian also highlighted Microsoft's Fairwater data-center projects in Wisconsin and Georgia, saying parts of the development appeared unfinished despite earlier statements that the Wisconsin facility was going live.
The chip-count concerns contrast with JPMorgan's view that Microsoft's AI infrastructure spending is translating into accelerating cloud growth. In fiscal 2026, Azure revenue surpassed $100 billion, up 41 percent for the full year, and management added 88 data centers during the fiscal year, including 31 in the latest quarter. Commercial remaining performance obligations grew 84 percent to $678 billion, and calendar-year 2026 capital expenditure is expected to land around $175 billion.
Microsoft has also been expanding its external compute partnerships. The company formally accepted IREN's Horizon 1 data center handoff at Childress, Texas — a 50-megawatt AI cloud deployment using direct-to-chip liquid cooling and Nvidia GB300 NVL72 systems — as part of its broader infrastructure push.
The tension between the Guardian's findings and JPMorgan's optimism reflects a broader debate about whether hyperscaler AI buildouts are running ahead of actual deployment capacity. Taiwan Semiconductor Manufacturing, the primary supplier of advanced AI accelerators, raised full-year 2026 revenue guidance to slightly above 40 percent year-over-year and hiked capital expenditure to $60 billion to $64 billion, with CEO C.C. Wei saying demand remains strong through 2029-2030.
For investors, the question is whether Microsoft's AI infrastructure is a capacity problem or a perception problem. If the chip gap is real, Azure's AI revenue growth could slow as delivery timelines slip. If it is a measurement dispute, the current selloff represents a buying opportunity at a valuation JPMorgan sees as undervalued. Microsoft shares trade at roughly 30 times forward earnings, a premium to the broader market but below the multiples commanded by some AI peers. The next catalyst will be Microsoft's fiscal first-quarter earnings, due in late October, when management is expected to provide updated data-center and chip deployment figures.
This article is for informational purposes only and does not constitute investment advice.