The AI Layoff List: A Sourcer's Q4 2026 Arbitrage Window

AI has been the leading cited reason for US job cuts four months running, a streak Challenger, Gray & Christmas confirmed before it broke in September. That streak produced something sourcers rarely get: a public list of named companies labeling which functions they consider automatable, which geographies absorbed the hit, and which AI-fluent roles they want to rehire. If you are not building saved searches off those filings right now, somebody else is.

Why the "AI-cited" filings are a free labeling exercise

Every AI-cited layoff announcement in 2026 is a public document telling you which function was deprioritized and which skills the company now treats as scarce. That is sourcing input, not news.

Through September 2026, AI has been cited in 120,136 announced cuts year to date, about 21% of all US job cuts. Since Challenger began tracking AI as a distinct reason in 2023, it has been cited in 184,538 cumulative announcements. Technology alone led all sectors with 149,023 cuts through July 2026, up 67% from 89,251 in the same period of 2025. Amazon, Oracle, Meta, and Microsoft alone account for almost 50,000 of the nearly 140,000 US tech jobs cut since the start of 2026.

120,136

The mechanism to understand: Challenger's own analysts have noted that naming AI in a filing can win over investors while pushing current employees away. The filings over-index on functions where AI framing plays well to the Street. Support, L1 engineering, and middle management get called out even when the real driver is cost. For a sourcer, that is a feature. The filings tell you which cohorts were deprioritized, which is exactly what you need to know to approach them.

The named companies, ranked by cohort usefulness

The useful filings are the ones that name a function, a headcount, and a date. Here are the anchor cuts worth building saved searches around.

Four of the five published a quotable artifact. Oracle's annual filing contains the line "the adoption and deployment of AI technologies across our operations have resulted, and may continue to result, in reductions to our workforce." Cloudflare CEO Matthew Prince wrote that "the vast majority of those we laid off last week were measurers." Monday.com co-founder Eran Zinman's LinkedIn memo is itself a sourceable artifact. GitLab CEO Bill Staples cited a "generational rebuild" of core infrastructure for 100x growth requirements. Those are the lines you paste into InMails to prove you read the room.

Cloudflare's "measurers" is a reusable template

The single most useful phrase in the 2026 wave is Matthew Prince's "measurers" line, because it tells you exactly which titles to pull every time a company issues an AI-first memo.

Prince defined measurers as middle management, finance, legal, internal auditing, and revenue recognition. That is a repeatable saved search. Any time a public tech company issues an "AI-first" memo, run a 30-day search for:

• Senior Manager through Director in Operations, FP&A, or Program Management
• 5 to 10 years of experience
• Any Revenue Operations, Deal Desk, or Revenue Recognition title
• Internal Audit, Compliance, or Legal Ops at the IC level
• Chief of Staff and Business Operations roles outside the exec suite

The mechanism: measurers are the easiest line item to defend cutting in an AI narrative, because their work maps cleanly to dashboards and agents in investor decks. Whether the agents actually work is irrelevant to the filing. What matters is that this cohort is now liquid, trained in SaaS operating cadences, and underpriced relative to their next employer's needs.

Describing those filters in Boolean takes 15 minutes per company. Describing them in English to Refolk takes one sentence, which is the gap worth closing when you are running the play across the whole Challenger list in a single week.

The support engineer arbitrage: 17,728 profiles, 42 with LLM skills

The displaced SaaS support engineer cohort is the single largest and most mispriced pool in the 2026 layoff wave, and almost none of them have the AI skills their former employers now say they want.

In Refolk's index of professional profiles, there are 17,728 US-based people with Support Engineer, Technical Support Engineer, or Customer Support Engineer titles. Of those, only 42 list LLM, Prompt Engineering, or LangChain skills. That is a 0.24% AI-fluency rate.

0.24%

Here is how the two cohorts actually trade:

1. The 42 AI-fluent profiles will be rehired at 1.5x comp by the same companies that just cut their orgs, or by their direct competitors. They are the smallest pool, and the one every AI-forward sourcer is already chasing.
2. The remaining 17,686 are not unemployable. They are trained in Zendesk, Salesforce Service Cloud, Jira, API debugging, and customer escalation. They are a direct fit for:
   • Non-SaaS enterprises building first-generation customer AI teams
   • Mid-market vertical SaaS companies that never had L2 depth
   • Dev tools companies that need developer-adjacent support, not AI research

That second group is where the arbitrage lives. The companies issuing AI-cited layoffs have told the market these people are now cost centers. The companies two sectors over have no such framing problem, and they will pay market rate for a trained support engineer with a real resume.

Monday.com's 630 cuts: a 22-person US GTM shortlist

Monday.com's July 22 layoff removed 20% of staff, about 630 people, with the US GTM cohort concentrated in two cities any competitor can source directly.

Refolk's index returns 22 current Monday.com AE, CSM, and SDR profiles in the US, with 6 in New York and 6 in Denver. Those two cities are the entire US GTM footprint worth sourcing. If you run a Monday.com competitor in the work management, CRM, or project ops space, your Q4 2026 pipeline from Monday.com alone is a two-city play, not a national one.

The structural read:

• New York (6): enterprise AEs and strategic CSMs, likely higher comp bands, longer sales cycles
• Denver (6): mid-market and SMB velocity reps, SDR leadership, lower base higher OTE
• Everything else: distributed and remote, harder to geo-target but easier to approach without office politics

Monday.com's filing cited $45M to $55M in restructuring charges, roughly $70K to $87K per departing employee in severance and transition. That is a signal about how long the cohort has before pressure to accept the next offer begins. The window is real but finite.

The "measurers" play by market

The reusable template works across any company that issues an AI-cited restructuring, but the role mix shifts by sub-sector. If you source into one of these markets, this is where the live demand sits right now.

Why the sourcing window closes in Q1 2027

The AI-cited layoff streak has already broken, which means the current displaced cohort is a bounded supply, not an ongoing one, and competing sourcers will converge on it within one or two quarters.

In September 2026, AI was cited for only 3,961 cuts, the fifth-most-cited reason that month. The four-month leading-reason streak ended. The narrative incentive to cite AI in filings is weakening as the "AI-first" framing starts to cost employer brand points instead of earning Street points. New AI-cited filings will slow, but the 184,538 cumulative cuts already announced since 2023 do not get re-released onto the market in waves. They are moving through pipelines right now.

Three forces compress the window:

1. Severance clocks: most severance packages run 8 to 16 weeks. The June and July 2026 cohorts are entering the "need the next job" phase in Q4.
2. Boomerang offers: companies that cut aggressively in Q2 2026 will quietly rehire a share of the cohort by Q2 2027, usually at lower levels. Those people disappear from your pipeline.
3. Competitor convergence: by Q1 2027, every sourcer running a Challenger-driven playbook will be chasing the same named companies. First-mover advantage on named cohorts is a 60 to 90 day window, not a year.

The filings are a signal of who was deprioritized, not who was replaced. That distinction is worth about 100,000 sourceable candidates.

The Meta exception: redeployment is not supply

Meta's 8,000-person cut is misleading on the open market because roughly 7,000 of those employees moved into new AI-focused roles internally, which means the external supply from Meta is closer to 1,000 than 8,000.

That matters for two reasons. First, Meta's displaced external cohort is lopsidedly non-technical and support-adjacent, not the ML and infra engineers the headline numbers imply. Second, the internal transfers are reportedly unhappy with the forced moves, which makes mid-level Meta AI-adjacent PMs and EMs a stronger passive-candidate target than the layoff pool itself. The flight-risk signal is inside the building, not outside it.

The same logic applies to Microsoft, where CFO Amy Hood said total headcount is expected to keep declining amid rising AI investment, but role eliminations are "not being replaced by AI" directly. Translation: the market cohort from Microsoft is smaller than the gross cut number suggests, and the retained headcount is the real poach target.

How to actually run this across the full list

Build one playbook, run it per company, by anchoring on the named function in each filing and the cities where the company has real office density.

For each company on the Challenger list:

1. Pull the filing or memo and extract the named function (support, measurers, infra, GTM, etc.)
2. Cross-reference layoffs.fyi, which has tracked more than 122,000 tech roles cut in 2026, for individual names and dates where community reporting fills in gaps the filing left vague
3. Build a saved search on title plus company plus 90-day tenure window
4. Layer the geography from company-level distribution, which tends to concentrate in 2 to 4 cities even for 5,000-person orgs
5. Score for AI fluency against LLM, LangChain, prompt engineering, MCP, and agent orchestration keywords

The multi-company execution is the hard part. Doing this once for Monday.com is a Tuesday afternoon. Doing it across every named company, with different filings, different functions, and different geographies, is a two-week project in the old stack. The input that collapses it is a sentence like "ex-GitLab infra engineers in North America hired before 2023, now showing GitHub activity on Terraform or Pulumi" and the output is a ranked list across LinkedIn, GitHub, and the open web.

The named company list is public. The arbitrage is in execution speed.

FAQ

Which AI-cited layoff cohort is the single best to source right now?

Cloudflare's "measurers" cohort, because Matthew Prince named the exact functions (middle management, finance, legal, internal auditing, revenue recognition) and the layoff landed in May 2026, which means severance clocks are running out in Q4. The 1,100-person cut was 20% of the company, concentrated in operational roles that transfer cleanly to any mid-market SaaS or fintech. Monday.com and GitLab are tighter and more targeted, but Cloudflare's pool is the deepest and most transferable.

Are displaced support engineers worth sourcing if they don't have LLM skills?

Yes, but not for the companies that cut them. In Refolk's index, 99.76% of US support engineers lack LLM, prompt engineering, or LangChain skills, and the companies issuing AI-cited layoffs have publicly labeled that majority as cost centers. The arbitrage is placing them into non-SaaS enterprises, vertical SaaS, and dev tools companies that value Zendesk, Service Cloud, and API debugging experience without needing the AI resume line.

How long does the Q4 2026 sourcing window stay open?

Probably 60 to 90 days per named company before competitor convergence erodes the advantage. The AI-cited leading-reason streak broke in September 2026, so new filings will slow, and severance timelines mean the June and July cohorts hit "need a job now" pressure in November and December. By Q1 2027, the named companies are either rehired, boomeranged, or absorbed into other pipelines. Treat the list as a bounded supply and prioritize the companies whose severance clocks are furthest along.

Is the Challenger data reliable enough to build a pipeline on?

Reliable enough for prioritization, not reliable enough for individual sourcing. Challenger counts announced cuts by stated reason, which captures the labeling exercise accurately but over-indexes on functions where AI framing plays well to investors. Andy Challenger, the firm's Chief Revenue Officer, is the quotable source on monthly shifts. Use Challenger to pick which companies to target, then use layoffs.fyi, LinkedIn tenure gaps, and GitHub activity patterns to find the actual individuals.