Call it AI. It's still a haircut.

Here’s the headline pattern.

[Tech company] cuts [round number] jobs. The memo says efficiency. The quote says AI. LinkedIn does its usual thing. Half the feed acts like the machines have arrived. The other half posts screenshots of how they shipped in one afternoon what used to take a week.

Both camps miss the point. The layoffs are real. The explanation usually isn’t.

What actually happened

Between roughly 2020 and early 2022, money was basically free. The Fed cut the target federal funds rate to 0.00%–0.25% on March 15, 2020 and didn’t start lifting it until March 17, 2022 (‘Open Market Operations’, Federal Reserve, updated 2026). That matters because rate regimes change behavior, not just spreadsheets.

Venture capital responded exactly how you’d expect. U.S. venture investment hit records in 2021 – PitchBook/NVCA put 2021 U.S. VC deal value at $358.2 billion (‘2026 NVCA Yearbook’, NVCA / PitchBook, 2026). That money had consequences. Companies hired ahead of demand because the incentive structure said to. Headcount became a growth proxy. In a lot of boardrooms, the number of employees was treated like evidence of momentum.

And the surge was real. Meta’s second round of 2023 cuts came after a hiring wave that had doubled its employee count versus 2020 (‘Meta to cut 10,000 jobs in second round of layoffs’, AP News, Mar. 14, 2023). Across big tech, the pattern was the same: massive payroll expansion during the cheap-money years, followed by a very public rediscovery of discipline.

The internal logic was simple enough. Capital was cheap. Talent was scarce. If you waited until you actually needed someone, somebody else would hire them first. So companies stocked up. That was the game. And for a while, it worked.

Then the regime changed. Rates rose. Capital got expensive. “Growth at all costs” stopped being strategy and started being a liability. The board question flipped from How many people did you hire? to Why the hell do you have this many people?

That’s the part companies can’t say out loud. Because “we overhired when capital was free and nobody stopped us” lands a little worse than “we’re reorganizing around AI.”

Enter AI

This is where the press release earns its salary.

If you need to cut 10% to 15% of staff, you need a story. The plain-English version is ugly: bad planning, bad discipline, changed conditions, too many people on the payroll. The AI version sounds visionary.

We are leveraging AI-driven efficiencies to optimize our workforce.

That sentence does a lot of work. It turns correction into strategy. It turns a rate-cycle hangover into innovation theater. And the playbook is older than the current buzzword.

Ghost jobs during layoffs are a documented phenomenon. CNBC, citing Resume Builder survey data, reported in 2024 that 4 in 10 companies said they had posted fake listings that year (‘Ghost jobs: Why fake job listings are on the rise’, CNBC, Aug. 22, 2024).

That matters because the function of the posting is not always hiring. Sometimes it is signaling. You cut 200 people, leave 50 shiny technical roles on the careers page, and point anyone asking questions toward “the transformation.” In 2021 and 2022, that flavor was often machine learning, automation, or digital transformation. In 2025 and 2026, it’s AI. Different label. Same move.

Zillow showed the template in 2022: lay off roughly 300 people, then emphasize that the company was still hiring in key technology-related roles (‘Zillow lays off 300 employees in latest workforce shift’, TechCrunch, Oct. 26, 2022). The point isn’t that every one of those roles was fake. The point is that the existence of future-facing technical reqs gives management a cleaner narrative: this is a pivot, not a correction.

AI is just the current wrapper on that same story. And by 2025, companies were getting less subtle about it. Workday cut 1,750 jobs – 8.5% of staff – while investing heavily in AI (‘Workday to cut 1,750 jobs in AI push’, AP News, Feb. 5, 2025). CrowdStrike announced 5% job cuts while saying AI is “reshaping every industry” (‘CrowdStrike announces 5% job cuts’, CNBC, May 7, 2025).

That’s the headline pattern now. And it’s worth reading with some suspicion.

The growth test

Jensen Huang, Nvidia’s CEO, has a cleaner frame for this than most consultants. His basic argument: if AI actually drives productivity and revenue growth, it should lead to more hiring, not less. In TIME’s 2025 profile, he put it plainly: “I’m fairly confident that AI will drive productivity, revenue growth, and therefore more hiring” (‘The Architects of AI Are TIME’s 2025 Person of the Year’, TIME, Dec. 2025).

If AI is truly transformative, the move is not to cut 100 people to 50 and hold revenue flat. The move is to keep the 100 and go build more. More product. More customers. More throughput. More market. That is the dividing line.

A company using AI as a growth lever looks different from a company using AI as a layoff alibi. The growth company keeps the team and points it at work that was previously too slow, too expensive, or too annoying to do. The cost-cutting company writes a memo, trims the org chart, and prays nobody asks what the AI is actually doing besides helping this quarter’s margin.

One of those is strategy. The other is cost accounting in a black turtleneck.

The question nobody is asking

This is the part that matters.

The AI productivity discourse is full of examples like this: we automated a report that used to take two weeks. Fine. Great. Standing ovation. Now ask the adult question: did anyone actually need the report?

If the answer is no – if that two-week report was a recurring artifact that existed because someone created it once and nobody ever killed it – then AI didn’t create value. It accelerated theater. That’s the trap in a lot of enterprise AI talk. People confuse doing the same low-value work faster with improving the business. Those are not the same thing.

The real gain is not “use AI to produce more internal sludge.” The real gain is to stop producing sludge and point the human at work that matters.

The actual upside shows up when AI makes new work economically possible. New analysis. New products. New customer segments. New service levels. Work that was previously too expensive, too slow, or too messy to attempt. That’s where revenue lives.

Most “AI efficiency” layoff stories are describing something else entirely. A headcount reduction, attributed to a technology, timed to a quarter when the board wanted margin.

What AI actually changes

AI is real. It is useful. It can compress cycle times, raise output per employee, and let smaller teams hit a scale that used to require larger ones. None of that automatically implies layoffs.

The companies that seem to understand AI best describe it as an expansion tool, not a shrink ray. Huang’s argument is one version of that. So is the market logic: if a tool lowers the cost of creating value, competent companies should create more value. Not less company. More company.

So the next time you see the headline – [Company] cuts [number], cites AI – ask one question:

What is the revenue target?

If the answer is flat, iffy, or missing, you are probably not looking at an AI transformation. You are looking at a balance sheet correction with better branding.