Cognition's 200 Windsurf Buyouts Won't Trigger WARN. Here's the Pool.

Three weeks after acquiring Windsurf, Cognition laid off 30 employees and offered voluntary buyouts to the roughly 200 who remained. The offer: nine months of salary, or stay and work 80+ hour weeks, six days a week, in-office. That is the most expensive silent layoff of 2025, and no layoff tracker will ever see it.

If you source AI-IDE engineers, this is your window. It closes fast.

Why WARN, layoff.fyi, and Crunchbase will all show zero

Because buyouts are voluntary separations, they sidestep every layoff-notification law and every scraper that depends on them. California WARN requires 60-day notice for 50+ involuntary layoffs at a single site. A nine-month package paired with an 80-hour, six-day-in-office alternative produces "voluntary" attrition on paper, even when the mechanism is engineered to push people out.

Here is what the trackers will and won't catch from the Cognition/Windsurf event:

• layoff.fyi: logs the 30 involuntary cuts. Misses the 200.
• State WARN databases: nothing filed. Threshold not met.
• Crunchbase's layoff tracker: recorded the 30 cuts and noted the 200 buyout offers, but will not record the eventual departures as layoffs.
• LinkedIn "Open to Work": will surface departures in staggered waves as the August 10 decision deadline plus notice periods play out over 60+ days.
• GitHub bio and personal-site updates: the earliest signal, often ahead of LinkedIn edits.

Scott Wu, Cognition's CEO, wrote in the email that the company doesn't "believe in work-life balance." That is not a culture statement. It is a filter, and the filter is doing exactly what filters do.

The pool this actually creates

Roughly 100 senior AI-IDE engineers will hit the market over the next 60 days, and they land on top of a US senior pool of only 78 people with the adjacent stack. That is the entire story.

Windsurf was not a small side project. The company shipped an IDE with 350+ enterprise customers and hundreds of thousands of daily active users before Google's $2.4B reverse-acquihire took CEO Varun Mohan, co-founder Douglas Chen, and the research leads. The remaining ~250-person team, the people who actually built and shipped the product, is the group Cognition inherited and is now pushing out. Jeff Wang, Windsurf's former head of business, was made interim CEO days before the Cognition deal and later called it "the wildest rollercoaster ride" of his career.

78

That 78-person figure is why this event matters more than the headcount suggests. A ~100-person cohort with production experience in agentic-IDE deployment at enterprise scale roughly doubles the sourceable senior pool for that specific stack. In most talent markets, an M&A event of this size is a rounding error. In this one, it is the market.

The table your leadership actually needs

The 29,233 number is the wide funnel. The 78 is the narrow one. The Windsurf cohort is the only event in 2025 that meaningfully shifts the 78.

Where these engineers will actually land

The realistic destinations are Cursor, Google DeepMind, Harvey, Linear, OpenAI, and Meta, in roughly that order of pull. If you are not one of those six, you are competing on speed and specificity, not brand.

Cursor (Anysphere) hit $500M ARR and is the obvious competitive landing spot for anyone who wants to keep building an AI-native IDE. But the more interesting buyer is Google DeepMind. Google took the founders and research leads in the acquihire but did not buy the IDE, the customer base, or the deployment expertise. It has to rebuild those components from scratch, and the cheapest way to do that is to quietly recruit from the exact buyout cohort it left behind three weeks ago. Watch for DeepMind recruiters reaching out to ex-Windsurf ICs in September and October.

In Refolk's index of professional profiles, the top current employers of US Software Engineers with TypeScript are Meta, OpenAI, Google DeepMind, Linear, and Harvey. Those are your competitors for this cohort, and every one of them has an internal referral pipeline into ex-Windsurf staff that started forming on August 6.

Buyout vs layoff sourcing: the operational difference

Layoff sourcing is a fire drill triggered by a public filing. Buyout sourcing is a slow bleed detected through weekly delta scans on tenure, bio, and commit signal. If you treat them the same, you will miss the buyout cohort every time.

The mechanics differ on four axes:

1. Trigger event: layoffs give you a date and a headcount. Buyouts give you neither. You have to infer both from tenure gaps and profile edits.
2. Signal medium: layoffs surface on WARN and layoff.fyi. Buyouts surface on GitHub commit cadence drops, LinkedIn tenure end-dates, and personal-site updates.
3. Timing: layoffs cluster on a single day. Buyouts stagger across 60+ days as individuals decide, negotiate, and serve notice.
4. Competition: everyone sees a layoff. Almost nobody notices a buyout cohort until it is already placed.

The right operational move for the Cognition/Windsurf event is a standing weekly scan on the ~200 profiles that show "Windsurf" or "Codeium" in current or recent experience, filtered on delta since August 5. That is a job for a query, not a spreadsheet, and it is the exact gap Refolk closes: you describe the cohort in plain English and get a ranked, dated shortlist that updates as profiles change.

Layoff sourcing is a fire drill. Buyout sourcing is a slow bleed you only catch with weekly delta scans.

The playbook this is the third instance of

Cognition/Windsurf is the third public run of a repeatable pattern: mega-deal takes the founders, remaining employees get buyout packages, quiet talent pool scatters. Sourcing leaders who don't build a watchlist for the next one will lose it too.

The pattern so far:

• Scale AI: Meta paid $14.8B, then Scale cut 200 people.
• Character.ai: Google paid $2.7B, founders joined Google, remaining employees got buyout packages.
• Windsurf/Cognition: Google paid $2.4B for the founders and research leads, Cognition bought the remainder for an industry-estimated ~$250M (a 92% discount on Windsurf's $82M revenue stream), then offered buyouts to ~200.

Sourcing leaders should build watchlists on the next likely reverse-acquihire targets - Perplexity, Poolside, Magic, and Cognition itself, which closed a $400M round at a $10.2B valuation two months after the Windsurf deal. The right time to start sourcing a reverse-acquihire cohort is before the acquihire, not after.

~$250M

What to do this week

Run three parallel scans and give each one a named owner on your team. If nobody owns the ex-Windsurf watchlist by Friday, you will read about the placements in a Cursor blog post in November.

1. Delta scan on ex-Windsurf/Codeium profiles: weekly. Filter on tenure end-date after July 15, GitHub commit signal in the last 30 days, US-based. Expect the list to grow across September and October as the 60-day trickle from the August 10 deadline plays out.
2. Adjacent-stack expansion: the 78-person senior React + agentic pool is your comparison baseline. Reach out to everyone in it now, before the ex-Windsurf cohort crowds the same inboxes.
3. Named-competitor tracking: watch DeepMind, Cursor, Harvey, and Linear recruiters' public activity. If a Cursor recruiter's LinkedIn shows five new "engineer" connections a week from ex-Windsurf accounts, you are already behind.

For the first scan in particular, plain-English querying beats Boolean strings. A search like "US engineers who list Windsurf or Codeium in their last two roles, ship TypeScript on GitHub, and updated their profile in the last 14 days" is the kind of query Refolk answers directly across GitHub, LinkedIn, and the open web without you rebuilding it every Monday morning.

The buyout deadline was August 10. The 60-day trickle started August 11. If your first outreach to this cohort goes out after September 15, you are not sourcing an acquisition, you are reading about one.

FAQ

Why don't Cognition's Windsurf buyouts trigger WARN Act filings?

WARN (federal and California) requires 60-day advance notice for involuntary layoffs above specific thresholds, typically 50+ employees at a single site. Buyouts are structured as voluntary separations, so they fall outside WARN entirely. The 30 involuntary Windsurf cuts also fell below the state threshold. That is why layoff.fyi, WARN scrapers, and even Crunchbase's layoff tracker will record at most the 30 cuts and the buyout offer itself, but never the actual exit wave that follows.

How many ex-Windsurf engineers will realistically hit the market?

The public facts are 30 laid off and ~200 offered buyouts, with a decision deadline of August 10 and a nine-month salary package. Assuming a mid-range 50% take rate (reasonable given the alternative is 80+ hour weeks, six days in-office), roughly 100 senior engineers scatter over the following 60 days. That is a derived estimate, not a filed number, which is precisely why layoff trackers won't show it.

What skill makes the Windsurf cohort disproportionately valuable?

Enterprise-scale agentic-IDE deployment experience. Windsurf shipped to 350+ enterprise customers and hundreds of thousands of daily active users before the Google deal. In Refolk's index, only 78 US senior Staff Engineers carry the full React plus agentic-tooling profile that maps to Windsurf's frontend stack, so a ~100-person cohort with production experience roughly doubles the sourceable senior pool for that specific work.

Which companies will hire the most ex-Windsurf engineers?

Cursor (Anysphere) is the obvious competitive destination at $500M ARR. Google DeepMind is the non-obvious one: it took the founders in the $2.4B acquihire but did not buy the IDE, customers, or deployment know-how, so it has to recruit that expertise back from the exact cohort it left behind. Harvey, Linear, OpenAI, and Meta round out the realistic landing zones based on top current employers of senior TypeScript engineers in Refolk's index.