If you are writing resumes right now, the January 2026 headlines have probably convinced you the market is imploding. The WARN data through September says something different, and the gap between those two stories is where a smarter Q4 strategy lives.
The WARN data says 2026 is quietly cooler than 2025
Through September 2026, LayoffAlert.org has tracked 3,245 WARN Act notices across 44 states, affecting 290,283 workers. At the same point in 2025, the same tracker showed 3,356 notices affecting 301,352 workers. 2026 is running 3.3% behind on notices and 3.7% behind on affected workers.
A second tracker, WARNact.io, shows an even steeper decline: roughly 163,273 affected workers so far in 2026 versus 261,814 last year, a 38% year-over-year drop. Neither number matches the "worst layoff year since the pandemic" framing you have been reading since January.
Down from 3,356 at the same point in 2025, across 44 states.
Here is the fuller picture the doom coverage keeps skipping:
| Metric | 2026 YTD | 2025 YTD | Delta |
|---|---|---|---|
| WARN notices (through early Sept) | 3,245 | 3,356 | -3.3% |
| WARN-affected workers | 290,283 | 301,352 | -3.7% |
| Challenger cuts, January only | 108,435 | 49,795 | +117.8% |
| Challenger cuts, February only | 48,307 | 172,017 | -71.9% |
| Challenger cuts, Jan-Feb combined | 156,742 | ~221,812 | -29.3% |
WARN is the closest thing to a hard count of who is actually losing a job. It is a legal filing required 60 days in advance for site closures or mass layoffs of 50 or more workers at a single site. And it is trending down.
January's 108,435 was a pull-forward, not a new baseline
Challenger, Gray & Christmas reported 108,435 U.S. job cuts in January 2026, up 118% from 49,795 in January 2025 and the highest January total since 2009 (when the number was 241,749). That single month powered every "record layoff year" headline you saw through Q1.
Then February happened. Employers announced 48,307 cuts in February 2026, down from 172,017 a year earlier, a 71.9% drop. The Jan-Feb combined total of 156,742 is the lowest since 2022 and roughly 29% below 2025's two-month pace.
The mechanism matters. Challenger's own read: "most of these plans were set at the end of 2025, signaling employers are less-than-optimistic about the outlook for 2026." Translation: January was a pull-forward. Companies front-loaded restructuring plans they had been sitting on. It was not a new run rate. If you have been writing every application from a defensive crouch built on January's number, you are pricing in a shock that already cleared.
What January was actually made of
The industry mix inside that 108,435 is the second thing the panic reading gets wrong:
- Contract Loss: 30,784 cuts, the single largest cited reason
- Market and Economic Conditions: 28,392 cuts
- Restructuring: 20,044 cuts
- AI as a cited reason: 7,624 cuts, just 7% of the January total
- Transportation: 31,243 cuts, the highest of any sector (largely UPS)
- Technology: 22,291 cuts, of which 16,000 were Amazon management delayering
AI is doing less of the direct killing than LinkedIn implies. Contract loss and macro conditions are doing most of it. That should change how you write bullets.
The tech reshuffle is real, and it is not showing up in WARN
Inside tech specifically, a violent internal reshuffle is happening that WARN largely misses. WARN triggers at 50 or more workers at a single site. Distributed layoffs at remote-heavy companies frequently fall through that filter, and buyouts do not trigger it at all.
The named 2026 cuts to keep in mind:
- Oracle: 21,000 jobs cut over the past year (~13% of its workforce), per its June 23 annual disclosure. The single biggest company impact of 2026.
- Amazon: roughly 30,000 jobs across October and January rounds, framed as management delayering.
- Microsoft: a first-ever buyout offered to up to 7% of employees, targeting workers at senior director level and below whose age plus tenure equals 70 or higher.
- Nike: 1,400 roles, with the majority in technology, per COO Venkatesh Alagirisamy.
- Cisco: just under 4,000 jobs in Q4, spread across offices in ways that often escape single-site WARN thresholds.
- Coinbase: 700 employees (~14% of staff), with CEO Brian Armstrong citing the market and AI-driven workflow changes.
WARN is calm because the cuts hitting knowledge workers are distributed, buyout-shaped, or below single-site thresholds. Challenger is noisy because it counts announcements regardless of geography or filing status. Both are true. If you are a laid-off engineer from any of the above, your resume is competing in a pool the WARN headline does not describe. The tech sector still leads all industries at 33,330 cuts YTD, followed by transportation at 31,702 and healthcare at 19,228.
The 23x supply problem is the real reason your resume is stalling
In Refolk's index of professional profiles, there are 347,874 U.S. software engineers versus 14,776 AI/ML engineers. That is roughly 23 conventional SWE profiles for every AI engineer profile in the market.
347,874 SWE profiles vs. 14,776 AI/ML engineer profiles.
This is the number that actually explains the ghosting. The layoff problem is smaller than the pool problem. When Oracle sheds 21,000 people and Amazon delayers 30,000, those engineers land in a market where the generalist "Software Engineer" title has 23 times the supply of the title employers are actively fighting over. Volume of applications does not fix that. Title differentiation on the resume does.
A laid-off backend engineer whose 2023 resume reads "Senior Software Engineer, Distributed Systems" is competing inside the 347,874-profile pool. The same person, honestly rewritten as "ML Platform Engineer" or "Applied AI Engineer" based on real projects they shipped (retrieval pipelines, eval harnesses, inference cost work), is competing inside the 14,776-profile pool. That is the arithmetic. This is the exact work Refolk takes off you: paste your history and the posting, and get a resume back that pulls the AI-adjacent work out of your last two years and puts it in the title line where ATS ranking actually looks.
What a fall 2026 resume rewrite should actually change
Three moves matter more than the rest, and they follow directly from the data above.
1. Delete the "AI-proof skills" framing
Only 7% of January cuts cited AI. The other 93% came from contract loss, macro conditions, and restructuring. If your cover letter opens by insisting you are "AI-native and future-proof," you are answering a question hiring managers largely are not asking. Instead, lead with contract-generation, revenue-retention, and cost-reduction outcomes tied to specific dollar figures. That is what the actual layoff reasons reward.
2. Retitle before you rewrite
If your last title was "Software Engineer II" and your last year of work included any of: RAG pipelines, eval frameworks, fine-tuning, agentic workflows, vector search, prompt infrastructure, inference cost optimization, or LLM observability, your resume title line should say so. The 23x supply gap is a title-matching problem before it is a skills problem. Refolk scores how well you actually fit a given posting and rewrites the title line and top bullets to match the posting's exact language, which is the difference between landing in the ~15K pool or the ~348K pool for a given ATS query.
3. Rotate one resume toward the sectors that are hiring
This is the contrarian data point almost nobody in tech-Twitter noticed. Andy Challenger, in a September statement: "Employers are hiring more than they were at this point last year, which bucks the trend we've seen since 2020. The demand is showing up in aerospace, energy, and manufacturing, work that happens on a floor rather than a screen." Overall hiring is up 25% year over year.
If you have transferable analytics, ops, embedded, controls, reliability, or supply-chain-adjacent experience, you should have a second resume version pointed at aerospace, energy, and manufacturing. Not instead of the tech search. In parallel.
The layoff problem is smaller than the pool problem, and the pool problem is a title-matching problem before it is a skills problem.
The ResumeBuilder 58% number, in context
A widely-cited ResumeBuilder survey found 58% of companies plan layoffs in 2026, citing AI adoption, economic uncertainty, and restructuring. That number has been used to argue the WARN data is misleading and the real cuts are still coming.
Read it against the actual filings. 58% of companies "planning" layoffs is a stated-intent survey, not a headcount. Some of those plans mean 10 people. Some mean 10,000. The WARN dataset (3,245 filings, 290,283 workers, 44 states) is the operational truth, and 2025 was itself a heavy filing year that 2026 is currently trailing.
The honest read for a Q4 job seeker: expect continued announcements, especially in tech, especially at large-cap software companies restructuring around AI product bets. Do not expect a broad economy-wide layoff wave. Write your resume for concentrated cuts in your specific corner, not for a general collapse.
What to do this week
A concrete checklist that follows from the data:
- Pull your current resume and count how many bullets tie to revenue, contract retention, or cost. If it is fewer than three, rewrite the top of your experience section. Contract loss drove 30,784 January cuts. Hiring managers are staffing against that.
- Rewrite your title line to match the highest-precision version of what you actually did last year. If AI-adjacent work is honest, use it. The 23x supply gap punishes generic titles.
- Build a second resume for aerospace, energy, or manufacturing if you have any transferable skill. Hiring is up 25% YoY there per Challenger.
- Stop applying with one master resume. Every posting should get a version tailored to its exact language, which is the friction Refolk was built to remove: paste the posting, get a version back that scores your fit and rewrites the bullets and cover letter to the posting's actual vocabulary.
- Ignore any headline that quotes only the January Challenger number. Ask what happened in February.
FAQ
Are WARN notices a reliable measure of total layoffs in 2026?
WARN is reliable for what it measures: mass layoffs and site closures of 50 or more workers at a single site with 60 days advance notice. It undercounts distributed and remote layoffs (Cisco spreading ~4,000 cuts across offices, Coinbase's 700 remote workers, Microsoft's buyout program) that fall below single-site thresholds. So the 3,245 notices / 290,283 workers figure through September 2026 is a floor on layoff activity for large concentrated events, not a ceiling on total knowledge-worker impact.
Why was January 2026 so much worse than February?
Challenger's own explanation: most of the January plans were set at the end of 2025, meaning January was a pull-forward of restructuring decisions rather than a new run rate. February dropped to 48,307 cuts, down 71.9% year over year, and the Jan-Feb combined total of 156,742 is actually the lowest two-month total since 2022. Reading January in isolation gave the entire year a doom framing the subsequent data has not supported.
Should I retitle my resume as "AI Engineer" if I have not shipped LLM features?
No. The 23x supply gap between generalist SWE profiles and AI engineer profiles in Refolk's index rewards honest title differentiation, not fabrication. If your last year included retrieval pipelines, evals, fine-tuning, agentic workflows, inference optimization, or LLM observability work, use language that reflects that. If it did not, lean into an adjacent honest specialty (platform, data infra, ML ops-adjacent) rather than claiming AI work that will collapse in a technical screen.
Which sectors should tech workers actually target in Q4 2026?
Aerospace, energy, and manufacturing, based on Andy Challenger's September statement that hiring in those sectors is up 25% year over year. Transferable skills include controls, embedded, reliability, analytics, supply-chain systems, and platform engineering. The move is not to abandon a tech search but to run a parallel one with a second resume tailored to those industries' vocabulary.