Microsoft just offered voluntary exits to about 60,000 employees. Buyouts feel gentler than layoffs, but they hand one of the most important workforce decisions to the people walking out the door.
Microsoft has offered voluntary buyouts to about 60,000 employees, roughly 27% of its workforce, across the U.S., Canada, the UK and India. The packages follow standard severance, scaled to years of service, and decisions are due around early December. Not every team is taking part. Each organization inside Microsoft decides whether to offer it, based on its own priorities and customer commitments. The company's chief people officer described the program as new and said Microsoft would learn from it.
The timing is telling. Microsoft is spending at a historic pace on AI infrastructure, with $41 billion in capital spending in a single quarter this year. The offer also follows about 4,800 layoffs in July, more than 15,000 in 2025, and an earlier voluntary retirement program this spring.
A buyout hands the decision to the employee
There's a strategic case for buyouts. People get to decide what's in their own best interest, on their own timeline, with support. That's more respectful than a list drawn up in a boardroom, and it usually does less damage to morale and reputation than a round of layoffs.
From a risk management perspective, though, a buyout puts a lot of autonomy in the hands of employees, and that autonomy doesn't always work in the organization's favour. Two risks stand out.
The first is the number. You can't control how many people say yes. Asked about expected uptake, a Microsoft spokesperson pointed to similar programs at other large tech companies, where 3% to 5% of eligible staff accepted. Applied to 60,000 people, that's roughly 1,800 to 3,000 departures. If you've built a budget around a savings target, you're now hoping the right number of people raise their hands.
The second is who. Economists call this adverse selection. When the people accepting an offer know more about their options than you do, the ones most likely to say yes are often the ones you'd least want to lose. Strong performers with in-demand skills can take the package and walk into a new job. People with fewer options tend to stay. If the people leaving sit close to customers or products, the gap shows up fast.
Microsoft's own design suggests it knows this. Letting each organization opt out is a tell: some teams clearly can't afford to lose anyone.
Follow the data centres
Microsoft frames the program as part of adapting the business while it invests heavily in AI. Read more plainly, this is about paying for data centres. Without them, AI can't scale. Without AI revenue, the data centres don't pay off. It's a chicken-and-egg problem, and payroll is one of the few big costs a company can actually move.
To be clear, a few thousand departures won't fund $41 billion a quarter on their own. The real signal is the shift in where the money goes: away from people and toward infrastructure, with investors watching for proof that costs elsewhere are under control. It also isn't a simple story of AI replacing workers. Reporting on the eligible teams includes some of Microsoft's own AI and Copilot groups.
What leaders should settle before offering one
Most mid-market firms and public agencies will never offer a buyout to 60,000 people. But voluntary exits and early retirement are common tools when budgets tighten, and the same risks apply at any scale. Before opening the door, settle three things.
1. Which roles deliver the greatest return. Not every role contributes equally. Know which ones drive revenue, service quality or mission delivery, and which you could lose without customers or the public noticing. Exclude the first group, or make acceptance subject to approval.
2. How work depends on whom. People don't work in isolation. One departure can stall three teams if that person holds a key approval, relationship or system. Map the dependencies between people and teams so you know where a single exit becomes a bottleneck.
3. What would be hard to rehire. Some skills take months to recruit and years to build: deep institutional knowledge, regulatory expertise, specialized technical talent. Rank roles by how hard they'd be to replace, and protect the ones at the top.
Then plan for the work, not just the headcount. Decide which tasks stop, which get automated and who picks up the rest, so the people who stay aren't left carrying everything.
A buyout is a reasonable tool. Just don't let it design your organization for you.
A buyout lets people choose what's best for them. Make sure you've decided what's best for the organization first.