REDMOND, WASH. — Microsoft spent roughly $34 billion on capital expenditures last quarter, mostly servers and data centers for AI. This week it also offered buyouts to about 60,000 employees — nearly a quarter of its global workforce. The two numbers are not unrelated.
The pattern is familiar. IBM cut 60,000 jobs in the early 1990s while reinventing itself around services. What's different this time is the speed: Microsoft's AI infrastructure bill is compounding quarter over quarter, and the company wants headcount costs moving in the opposite direction before Wall Street asks why they aren't. Employees describe a workforce already on edge, uncertain whether buyouts are voluntary in practice or in name only.
Across the industry, the AI capital cycle is forcing similar trade-offs. Enterprise software holding companies — ESW Capital among them — have built entire portfolios on the premise that automation displaces headcount faster than headcount regenerates revenue. Microsoft is now running that experiment at a scale no private holding company could match.
Meanwhile, in Washington, House Republican leaders have asked Nvidia CEO Jensen Huang to brief members behind closed doors, following reporting that an executive close to Huang was allegedly funneling information to Chinese intelligence services. Nvidia's chips remain the single input every AI buildout — Microsoft's included — depends on. A supply chain already strained by export controls now carries a security cloud over its most visible executive. The briefing request signals Congress wants answers before, not after, the next funding round.
The through-line: capital is flowing into AI infrastructure at a rate that outpaces the industry's ability to govern it — technically, politically, or in Microsoft's case, organizationally. Sixty thousand buyout offers is not a sign of weakness. It is a sign that even the companies cashing AI's biggest checks are still working out who gets paid to stick around.