Look at the tenure column on most tech resumes and you’ll see the same shape. One long stint at one company, two if you’re lucky. Then a lot of short ones.

My journey is the same. I’ve been in the full-time workforce 32 years now, fourteen of them were at Microsoft. But the next longest is five and my current tenure just clicked over three, which makes it third most on my list. That leaves a lot of ~2 year roles, which in my defense has some partial attribution to moving around a bit in the 90s and early 2000’s.

But I suspect this tenure shape is quite common in the tech and security industries. Outliers exist for sure, I know plenty of people who have 20+ year journeys at one company.

This pattern is older than AI, but what AI and enterprise behavior are doing right now is a bit different.

Leaving was the promotion path

The old logic often required leaving in order to accelerate career ladder growth. You get hired at a level and you do the job well. Then you wait for the next level, which takes years if it comes at all.

Meanwhile, another company will pay market rate for the same skills. The sad reality is that most companies are better at hiring a stranger than promoting the known quantity.

So people leave. Some come back, often a level above where they would have been at if they had stayed. I don’t think this was a system design, it just became that way over time.

The cost of this behavior is paid by both sides. There is an upside on the human journey, because you learn new things and different workflows, and those make you better and more valuable on return, but there is also the cost of change, from benefits to stock resets and just life change. And the company gets a perceived accelerated growth of the person, but also they had recruitment and delivery impacts.

Binge and purge

Something different is happening at large tech companies. They shrink some teams through layoffs and buyouts. They keep hiring for AI and cloud. Both can be and are true at the same time.

This isn’t a significant contradiction, at least not yet. It’s a reshaping. The mix is changing, with roles no longer required or needed at the same volume being removed, and new different ones tied to AI.

Microsoft has made the point, and sharply. Following their first Voluntary Retirement option back in April, on October 26 it opened a Voluntary Exit Offer to about 60,000 employees. That is roughly 27% of its global workforce, across the US, UK, Canada, and India. Staff have to decide by early December and leave on December 31. And to emphasize the point above, data center, cybersecurity, and quantum teams are excluded. (International Finance)

Google has run the same play on a team by team basis. It has offered exits since 2025 and recently extended them to Canada, France, Germany, Japan, and Switzerland. It has said it will keep hiring in AI and cloud. (Yahoo Finance) Meta took the harder route. It announced about 8,000 job cuts in April and left 6,000 open roles unfilled … which makes me wonder whether they could not, or would not, retrain existing staff into those roles.

The driver is the AI bill. Alphabet expects 2026 capital spending of $195 billion to $205 billion, more than double 2025’s $91.4 billion. Its Q3 capex exceeded operating cash flow. Critics warn that buyouts can lose key staff and may not move costs much against budgets that size.

Buyouts are only one tool, and the cuts behind them are bigger. TrueUp’s 2026 layoff tracker lists the year’s largest tech and software cuts, and several tie them to AI.

Company Reported cut Context
Oracle 21,000 to 30,000 Operational shift toward AI deployment
Amazon ~16,000 to 17,000+ E-commerce, cloud, devices, and warehouses
Microsoft Buyouts for up to 60,000 Plus 4,800+ cuts in divisions like Xbox and Windows
Intuit ~3,000 (17%) Embedding AI deeper into its workflows
Atlassian ~1,600 (10%) Savings redirected to AI and enterprise sales
Cloudflare 1,100+ Restructuring tied to AI
Canva 1,271 (22%) Higher-than-planned AI running costs

Not every cut is about AI, but if you read the data, many are. I call this the “binge and purge” model. Hire fast when a bet looks good, cut fast when the bet changes. Repeat.

All of this puts tenure in an interesting spot. It used to be a choice the employee made. Now it is often a decision made for them by the employer. Years of service rarely protect a role the company has decided is no longer in the plan.

Microsoft’s two programs this year show the shift. April’s Voluntary Retirement Program required age plus years of service to total at least 70. About 2,600 of 8,750 eligible employees took it. October’s offer has no age or tenure test at all. Tenure used to be the entry ticket. This time it isn’t. Some people will choose to leave. Some will be left alone. But increasingly clear is that some are paid to go.

The AI wrinkle

Last week I ran an experiment on pmmkickstarter.com. I gave an AI assistant five prompts, typed the way a brand-new PMM would type them in a blank thread and no follow-ups. Then I graded the output like a new hire’s first week.

The results beat my expectations. The work read like something approximating competent product marketing. The full run is in Help, I’m the PMM now.

It had limits for sure. Early answers were generic until the tool saw the company homepage, an unvalidated positioning statement became the base of the launch plan, and advice to talk to real customers appeared, but buried in footnotes. Context was the bottleneck.

Now connect that to tenure. Part of what a long-tenured person offers is accumulated context and a feel for how the work gets done. A tool that drafts credible execution shrinks that gap but it doesn’t erase it, because the tool needs context to get specific.

I foresee two things: People who adopt the tools can protect their roles by doing more in the same hours, and they can also reach for roles that sat out of range before. If that holds, moving gets easier and tenure could shrink further. Where context is the hard part, long tenure may gain value instead.

What to do with this

Tenure used to tell a story about loyalty or ambition. Now it also tells a story about timing. Three things I think may help:

  1. Treat tenure as a decision. Once a year, ask what this role teaches you that another one can’t. If the answer is nothing, that’s helpful information, but not a decision.
  2. Build context that travels. Tools draft well, but they struggle with company-specific knowledge, customer conversations, and judgment. Those are what you grow in a role. Test the tools on a task you know really well, and grade the result to learn where you still matter.
  3. Separate the role from yourself. If you’re caught in a purge, the company made a call about a budget and a bet. It did not measure your worth.

Look at your own tenure column. Which entries did you choose, and which were chosen for you?

Sources