Data as of 28 September 2026. Every figure links to its source.
In February 2026, Block cut its workforce from more than 10,000 to under 6,000 and pointed to AI: "Intelligence tools have changed what it means to build and run a company." Headlines like that make it feel as if AI coding tools caused the tech layoff wave.
The timeline says otherwise, or at least says it is more complicated. The worst year for tech layoffs was 2023, when the most advanced coding AI most developers had was autocomplete. Here is how the two stories actually lined up, year by year.
2021–2022: AI arrives as autocomplete
GitHub Copilot, the first mainstream AI coding assistant, became generally available on 21 June 2022 for $10 a month. It suggested the next lines of code inside your editor. It did not run tasks or open pull requests on its own.
ChatGPT followed on 30 November 2022. Most of the research on AI and jobs uses that date as the starting line.
Layoffs.fyi counted 164,969 tech layoffs in 2022, as interest rates rose and pandemic-era demand cooled.
2023: the peak, with no AI agents in sight
Tech layoffs hit 262,735 in 2023, 59% more than 2022. It is still the record in this cycle. Read the reasons the biggest companies gave at the time:
- Microsoft (10,000 jobs): "macroeconomic conditions and changing customer priorities".
- Alphabet (about 12,000): "we hired for a different economic reality than the one we face today".
- Amazon (just over 18,000): "the uncertain economy and that we've hired rapidly over the last several years".
None of them blamed AI. The explanation was over-hiring during the pandemic boom, followed by higher interest rates. The one notable AI mention came from IBM, whose CEO said in May 2023 he "could easily see 30%" of about 26,000 back-office roles replaced by AI over five years. That was a slowdown in hiring for those roles, not a layoff.
2024: fewer layoffs, and the first "AI software engineer" demo
Layoffs.fyi counted 152,922 tech layoffs across 551 companies in 2024, well below 2023. In March, Cognition introduced Devin, billed as "the first AI software engineer". It got a lot of attention, but agent-style tools were not yet in everyday use at most companies.
2025: coding agents arrive, and layoffs keep falling
This is the year AI coding moved from suggesting lines to doing whole tasks:
| Date | Launch |
|---|---|
| 16 May 2025 | OpenAI Codex, a cloud software-engineering agent (research preview) |
| 22 May 2025 | Claude Code generally available, alongside Claude Opus 4 and Sonnet 4 |
| 25 June 2025 | Google Gemini CLI, open source |
Yet tech layoffs fell again, to 122,549 on Layoffs.fyi. Challenger, Gray & Christmas, which records the reason employers give, counted 54,836 US job cuts attributed to AI in 2025. The best-known cases were outside engineering: Salesforce cut customer support from about 9,000 to 5,000, and Chegg cut 45% of staff, citing "the new realities of AI" and less traffic from search.
2026: AI becomes the stated reason
This is where the story changes. Through August 2026, Challenger counted:
- 155,126 Technology-sector job cuts, up 52% year over year and 29% of all announced cuts.
- 116,175 cuts attributed to AI, about 22% of the total and already more than double 2025's full year.
- AI was the leading monthly reason from March to July, before dropping to fourth place in August with 3,462 cuts.
Block is the clearest case of a company naming the tools. Fortune reported that Block's leaders tested AI over the 2025 holidays and recognised that models like Anthropic's Opus 4.6 and OpenAI's Codex 5.3 were more powerful than before. Atlassian followed in March, cutting 1,600 jobs and citing the AI shift.
But even Block's story is mixed. The day after the announcement, Jack Dorsey posted: "yes we over-hired during covid because i incorrectly built 2 separate company structures (square & cash app) rather than 1". And Challenger's AI figures are self-reported: they record the reason a company gives, not a verified cause.
What the research says
| Study | Finding | How strong |
|---|---|---|
| Stanford Digital Economy Lab (Aug 2026) | Employment of 22–25-year-olds in AI-exposed jobs is 19% below trend. Mostly fewer hires, not more firings. No comparable gap for experienced workers. | Strongest evidence of an effect, but the authors call it descriptive, not causal |
| Yale Budget Lab (Oct 2025) | No "discernible disruption" to the broader labour market 33 months after ChatGPT | Economy-wide; finds little effect so far |
| Indeed Hiring Lab (Jul 2026) | AI-exposed jobs saw bigger posting declines, but the declines began before ChatGPT. Since May 2025 those same jobs have led the rebound. | Correlational; Indeed says correlation does not imply causation |
So, did AI cause the tech layoffs?
- The 2023 peak: no. It came before coding agents existed, and the companies said it was the economy and over-hiring.
- 2024–2025: mostly no. Layoffs fell while the tools got much better.
- 2026: partly, by companies' own account. AI is now the most-cited reason, but that is what employers say, and several also admit to over-hiring or are moving money into AI data centres.
- The clearer effect is on hiring. Young workers in AI-exposed jobs are being hired less. That does not show up in any layoff count. We broke down the entry-level numbers here.
If you use these tools every day, as I do, the result is not surprising: they make experienced developers faster, and that changes how many new people a team needs to hire. If you pay for them, our guide to what to know before you pay for AI covers how usage limits actually work.
If you are dealing with a layoff right now, work out your savings runway first.
Rebuild the chart yourself
Copy the numbers below into our chart builder, pick a bar or line chart, and export it as a PNG. It runs in your browser with no signup.
year,tech_layoffs
2022,164969
2023,262735
2024,152922
2025,122549Add a row when Layoffs.fyi publishes its full-year 2026 figure and see which way the line goes.