AI Is the Reason on the Layoff Memo Now. Read It Closely.

The layoff memo used to hide the reason. Now it leads with it. Monday.com, Salesforce, Coinbase, Block, all pointing at AI on the way out the door. Some mean it. Some are dressing up an ordinary cut in a fashionable word. Either way, the story your company tells about headcount just changed, and you own a piece of it. Let me catch you up.

The word on the memo is doing two jobs

Start with the count. U.S. tech companies have cut nearly 140,000 jobs this year, with Amazon, Oracle, Meta, and Microsoft alone behind almost 50,000 of them. Monday.com cut about 20% of staff, over 600 people, and called it a shift to a "leaner, more focused operating model." Co-founder Eran Zinman told staff the move "was not made to reduce costs or replace people with AI." Read that twice. The denial is doing work.

Here is the tell. The Financial Times found companies that named AI as a factor underperformed the Nasdaq by nearly 10% in the 30 trading days after their announcements. The market is not fully buying the story. AI on a layoff memo is partly a real operating change and partly a costume for a plain cost cut. Your job is to know which one you are running, and to say so plainly.

Headcount is not shrinking, it is moving

The cleaner read is that roles are getting reshuffled, not deleted. Meta laid off about 8,000 people while moving roughly 7,000 into new AI roles that, per the reporting, they hate. IBM is tripling entry-level hiring for AI and hybrid-cloud work even as it cuts elsewhere. Anthropic and OpenAI are absorbing talent shed across the industry.

John Cutler names the pattern well. A role, he writes, is "only a temporary bundle inside the system." Capabilities do not vanish when a title does. They specialize, diffuse to more people, get pulled into a central team, get baked into tools, or get bundled back together into broader jobs. Right now all of that is happening at once. When Coinbase floated "one-person teams" combining engineering, design, and product, that is rebundling, not magic.

So the honest question for your org is not "how many can I cut." It is which capabilities are moving, in which direction, and whether you are steering that or just watching it happen.

The output looks fine right up until it doesn't

Here is where the fashionable cut turns into a real problem. Cutler points to the MIT Work of the Future warning: strip out the middle tasks and you can keep the output while destroying the skills, networks, and career paths that produced it. The chart still looks right. The judgment behind it is gone.

Cloudflare's CEO said "the vast majority of those we laid off last week were measurers", meaning middle managers, finance, legal, auditing. Maybe. But Cutler's sharpest point is that people confuse a familiar-looking artifact for the work behind it. A screen, a chart, a roadmap. "Doesn't this look like something I could do?" AI reproduces the visible artifacts of expertise long before it reproduces the judgment that made them worth anything.

The academic beat says the same thing from a different angle. As production gets cheap, value moves to the things AI still can't do: choosing the right problem, knowing when the machine is confidently wrong, and mentoring the next person into that same judgment. Cut the apprenticeship layer and you save money this year and lose your bench next year.

Your team's actual skill is the hedge

The leaders who come out ahead are not the ones with the best AI slide. They are the ones whose people have used the tools enough to smell a bad output. That skill is cheap to build now. Phil Morton's advice is to have your people build a small side project end to end, not to prompt a toy in Lovable, but to work "closer to the metal" with Claude Code or Cursor and learn how software really ships.

His numbers are worth repeating in a planning meeting. He built a working app in two weeks as a busy parent, on a £20-a-month subscription and free tiers of GitHub, Vercel, and Supabase. "Nothing compares to what you learn by making." A designer who has shipped one thing knows what an AI tool is quietly getting wrong. A designer who has only watched a demo does not.

The deep cut

Watch the denials. When a CEO says the cut is "not to replace people with AI" or Cisco's CFO says it was "really not a savings-driven restructure," they are managing a story, not describing a system. Your own leadership will want you to reach for the same phrasing in your next planning cycle. Don't just borrow it.

Before you sign off on a headcount plan that names AI, do one concrete thing: for each role you're cutting or merging, write down which of Cutler's motions it actually is. Is the capability moving into a tool, moving to another person, or genuinely gone? If it's "gone" and you can't say where the judgment now lives, you are not restructuring. You are eating your own bench and calling it strategy. That single audit is what separates a real operating change from a cut in a costume, and it's the thing you can bring to your review that no one else in the room will have done.

Three questions for your team

1. For every role we plan to cut or merge, can we name where its judgment goes next, into a tool, a person, or nowhere? If the answer is "nowhere," what breaks in a year?
2. Which of our people have actually shipped something with AI tools, and which have only watched a demo? What's our cheapest path to fix that gap in the next quarter?
3. If we name AI in our own headcount story, is it true, and can we defend it to the team we're keeping without sounding like we're hiding a cost cut?