Monday.com Is the 21st AI-Cited Layoff of 2026. The US Pool Is 365.

TechCrunch's July 25, 2026 running list added Monday.com as the 21st major tech company to explicitly blame AI for 2026 layoffs, joining Coinbase, PayPal, Salesforce, Block, GitLab, Cloudflare, and Microsoft. Read the announcements and you'd think engineering headcount is collapsing. Look at who these same companies are actually trying to hire and a different story shows up: they're not shrinking, they're bar-raising, and the pool they're bar-raising into is a rounding error.

What Monday.com actually announced

Monday.com became the 21st major tech company in 2026 to explicitly cite AI as the driver of workforce reductions, and the restructuring will cost the company between $45 million and $55 million in charges. That framing matters because Monday.com is not in crisis. Neither is anyone else on the list.

The pattern across 2026 is loud and consistent:

• Cloudflare cut ~1,100 people (about 20% of its workforce) in the same quarter it reported $639.8M in revenue, up 34% year-over-year, the highest single quarter in company history.
• Block cut 4,000 jobs, nearly half its workforce, dropping from over 10,000 to under 6,000. Jack Dorsey framed it on X as a shift to "smaller and flatter teams" powered by internal AI.
• Salesforce cut 5,000 roles; Marc Benioff said AI agents now handle roughly 50% of customer interactions.
• GitLab laid off ~350 workers (about 14% of staff) to fund AI infrastructure and, per CEO Bill Staples, a "generational rebuild" for 100x growth.

None of these are cost-cuts in the classic sense. Cloudflare firing 1,100 people at a record-revenue peak is not a headcount reduction. It's a personnel swap dressed as a strategy update.

The 21 AI-cited layoffs are one story, told 21 ways

The "AI layoffs 2026" label is doing enormous PR work, and the trackers underneath it do not agree on what they're counting. Read the fine print and the same year looks like two completely different labor markets.

Here's the definitional gap the coverage keeps glossing:

• Layoffs.fyi: a record 78% of companies have blamed a need to refocus around AI for letting people go this year, with more than 122,000 tech roles cut so far in 2026.
• SkillSyncer: 54% of 2026 layoff events (173 of 322) cite AI, automation, or ML as a contributing factor, impacting ~170,945 workers.
• Challenger: AI was cited in 38,579 US job cuts in May 2026 alone, 40% of that month's total and the highest since Challenger started tracking AI as a reason in 2023.
• TechJack Solutions: classifies only 7.8% of tracked events as "AI-Direct," with 83.8% landing in "Business Cycle."

Same events. A 48-point gap between "AI is eating jobs" and "AI is a convenient narrative." Markets seem to know: companies citing AI as a factor in job cuts have underperformed the Nasdaq by nearly 10% in the 30 trading days following their announcements.

48pt

The mechanism is not mysterious. Saying "AI made us faster" is a cleaner reputational story than "we over-hired in 2021 and interest rates caught us." It also front-runs an internal message to the survivors: you now do more, with tools, or you're next. The layoff notice is half severance, half performance memo.

The pool they're actually hiring is comically small

Every one of the 21 layoff companies is quietly fishing in the same three-digit pond of publicly identifiable AI-native engineers. In Refolk's index of professional profiles, the entire US supply of software engineers who signal daily fluency with the leading agentic coding tools is roughly 365 people.

Here is the raw shape of it, US only, current title, straight from Refolk's index:

Set that 365 against the ~170,945 workers impacted by AI-cited layoff events in SkillSyncer's tracker and the ratio is roughly 468 to 1. For every one publicly identifiable "AI-native" US SWE, 468 workers have been laid off under an AI banner. The narrative is loud. The talent replacing them is not.

365

That pool is where compensation for verified AI-native ICs has decoupled from generalist SWE bands. A senior engineer who can prove day-to-day Cursor and Claude Code output is not being priced against the median staff engineer. They're being priced against the last competing offer, and there are not many of those to compare against.

The concentration is at the layoff companies themselves

The uncomfortable finding for anyone reading the layoff list as a poach list: the same companies making the loudest AI-cut announcements are also the largest current employers of scarce AI-native ICs. Firing at the front door does not fix a leak at the back door.

Top US employers of software engineers who list "Claude Code" on their profile, per Refolk's index:

• Meta (6)
• Apple (5)
• Microsoft (2)

Top employers for the "Cursor" cohort are more diffuse: Uber, Intuit, Microsoft, Chewy, Indeed, T-Mobile, one profile each, no dominant hub. That tells you two things. First, Claude Code adoption is concentrating at the FAANG tier where Anthropic partnerships and internal access are seeded early. Second, Cursor adoption has already leaked into mainstream enterprise, but not densely.

Firing at the front door does not fix a leak at the back door.

If you run recruiting at any of the 21 layoff-list companies, your retention risk is materially larger than your hiring risk. Losing two of the six Meta ICs who publicly signal Claude Code fluency is a 33% dent in a cohort you cannot backfill from LinkedIn in a quarter. This is where sourcing tooling stops being a productivity toy and starts being an intelligence layer. Refolk exists so you can ask, in plain English, "who at Meta lists Claude Code and has committed to an MCP server in the last 90 days," and get a real answer instead of a Boolean string that returns nine profiles and 4,000 keyword false positives.

Why LinkedIn skills undercount the real pool by an order of magnitude

The 365 number is the visible pool, not the actual pool, and the gap between them is the entire sourcing opportunity for 2026. Most Cursor and Claude Code daily drivers never edit their LinkedIn skills section. They ship instead.

The behavioral signals that identify real usage live outside profile fields:

1. Authors of public CLAUDE.md files in GitHub repos. This is the single cleanest proxy for a Claude Code daily driver.
2. Contributors to MCP (Model Context Protocol) servers. MCP is Anthropic's plugin protocol; contributing to a server means you've moved past "user" into "extender."
3. Commits to .cursorrules in project repos, which signals Cursor is not just installed but tuned to a codebase.
4. Members of the Cursor Discord, Anthropic's Claude Developers Discord, and the Latent Space / Interconnects reader base, which correlate with power-user status more reliably than any resume line.
5. High commit velocity paired with AI-generated commit patterns, the "vibes match" that any senior reviewer can spot in thirty seconds.

If profile-based signaling undercounts real usage by 10x to 50x (a conservative range given how few engineers touch their LinkedIn once employed), the true US AI-native SWE pool is somewhere between 3,650 and 18,250. Still small. Still concentrated. But findable, if you're looking at behavior instead of self-declared skills.

The reason plain-English queries matter here is that the sourcing question is compound. "Cursor power users recruiting" is not one filter; it's five, joined by AND, across three data sources. Boolean strings collapse under that load. This is the exact gap Refolk closes: you describe the person in plain English (skill, employer, tenure, artifact) and get a ranked shortlist that respects all five constraints at once.

What to actually do in Q3 and Q4 2026

Stop reading the AI-cited layoff list as a source of candidates and start reading it as a heat map of who is competing for the same 365 profiles. Then rewire sourcing around behavioral signals, not resume keywords.

Five concrete moves:

1. Build a retention watchlist before a hiring one. If you're at Meta, Apple, or Microsoft, the six-plus AI-native ICs on your team are the poach target for every other layoff-list CEO writing "smaller and flatter teams" tweets this quarter.
2. Source from artifacts, not titles. CLAUDE.md authorship, MCP server contributions, and .cursorrules history are more predictive than any "Claude Code engineers hiring" keyword search.
3. Ignore the 122,000 laid-off number for AI-native roles. The overlap between "cut in 2026" and "ships with Cursor daily" is thin. The laid-off pool is real and hire-able, but it's the pool for platform, infra, and non-agentic IC roles, not the top of your AI-native funnel.
4. Watch the AI-cited layoff underperformance signal. If those companies underperform the Nasdaq by ~10% for 30 days after announcing, expect equity refresh gaps and morale dents on months 45 to 90 post-cut. That's your outbound window.
5. Price AI-native ICs off the pool, not the band. 365 people means the market clearing price is set by the last competing offer, not by your comp bands from 2024.

For the highest-leverage version of move 2, Refolk indexes GitHub, LinkedIn, and the open web together, so a query like "US staff SWE who committed to an MCP server this quarter and previously worked at Anthropic or Cursor" returns one list, not three tabs of dedupe work. That is the operating layer sourcing needs when the addressable pool is measured in hundreds.

FAQ

How many US engineers actually use Cursor or Claude Code daily?

The publicly identifiable pool in Refolk's index is 144 US SWEs (SWE / Senior / Staff) listing Cursor and 221 listing Claude Code, totaling roughly 365 with likely small overlap. Broadened to anyone in the US with either skill in any role, it's 3,682. The real day-to-day user base is almost certainly 10x to 50x larger, but invisible without behavioral proxies like GitHub commits, CLAUDE.md authorship, and MCP contributions.

Why do AI layoff trackers disagree so much?

Because "AI-cited" and "AI-caused" are different questions and the trackers pick different ones. TechJack's stricter methodology calls only 7.8% of 2026 events "AI-Direct," while looser trackers put AI-related events above 54%. That 48-point gap is a definitional choice, and it's why the same year's layoffs can be read as either a labor apocalypse or a rebranding exercise.

Are laid-off engineers a good source of AI-native hires?

Mostly no, for AI-native roles specifically. The overlap between the ~170,945 workers impacted by AI-cited layoff events and the ~365 US SWEs who publicly signal Cursor or Claude Code fluency is small. The laid-off cohort is a strong pool for infrastructure, platform, and traditional SWE roles, but the AI-native top-of-funnel still needs to be sourced from behavioral signals, not layoff lists.

What's the single best signal that someone is a real Cursor or Claude Code power user?

A recent CLAUDE.md commit or .cursorrules file authored on GitHub, ideally in the last 90 days. Skills sections lie or lag; commit history doesn't. Pair that with membership in Cursor's Discord or Anthropic's Claude Developers Discord and you have a signal set that outperforms any LinkedIn Boolean search for the same cohort by a wide margin.