Microsoft's AI binge is under investor scrutiny
Investors are fretting over Microsoft’s massive AI spending and OpenAI reliance — concerns include a $120B AI capex plan, partner concentration risk, and slow Copilot seat uptake. That skepticism means healthtech buyers may demand clearer ROI and multi‑cloud escape hatches when evaluating AI-enabled vendor stacks. (simplywall.st) (markets.financialcontent.com)
Microsoft’s share-price wobble this week tracked specific disclosure metrics, with Simply Wall St citing a share level near $372.74 and a 6.7% one-week decline and about 21.2% year‑to‑date drop. (simplywall.st) Management’s capital plan signals scale: Microsoft has guided to a roughly $120 billion annual capex run‑rate for fiscal 2026 to build AI infrastructure including the “Fairwater” site in Wisconsin and global data‑center expansion. (markets.financialcontent.com) Company filings and reporting show concentration pressure — Simply Wall St reports about 45% of Microsoft’s commercial backlog is tied to OpenAI — and leadership moves have shifted Copilot engineering toward internal model work while Mustafa Suleyman’s role pivots to broader superintelligence initiatives. (simplywall.st) Product metrics add to scrutiny: Microsoft disclosed Microsoft 365 Copilot has reached 15 million paid seats, which represents only a low single‑digit share of the roughly 450 million M365 seat base, a figure some analysts described as disappointing relative to expectations. (computerworld.com) Strategic responses are visible — Microsoft has been reported to expand partnerships (including healthcare content deals) as it reduces exclusivity around OpenAI and OpenAI itself has signaled multicloud moves, altering the distribution and dependency calculus. (marketscreener.com) Market and IT leadership commentary links the capex/debt surge to buyer behavior: CNBC and CIO reporting note hyperscaler AI spending is increasingly debt‑funded and CIOs are re‑evaluating cloud lock‑in and ROI, driving enterprise demand for clearer ROI metrics and multicloud escape options when assessing AI vendor stacks. (cnbc.com)
