September's 199% Layoff Spike Is a Menlo Park Sourcing Map

The AI Career Hub tracker just posted a number that should reshape your next 30 days of outbound: September 2026 tech layoff events rose 199% over August, and one sub-sector, Major Internet, accounted for 77% of the month, roughly 4,400 workers. Zero announced hires posted in the same feed the same month. If you sell sourcing, that ratio is the whole game.

What actually happened in September 2026

September 2026 was not a broad tech downturn. It was a concentrated dislocation inside Major Internet (Meta, Google, and their peers), and the concentration is the sourcing map.

The AI Career Hub feed logged a 199% month-over-month jump in layoff events, with Major Internet alone producing 77% of the count. Cross-checking against the two independent trackers gives the pace:

• TrueUp: 548 layoff events in 2026 to date, 176,306 workers impacted, 760 per day.
• Skillsyncer: 383 events, 210,741 workers, 811 per day, as of September 17.
• 2025 baseline: 674 per day. The 2026 pace is roughly 13% to 20% higher, depending on which tracker you take.
• Layoffs.fyi's first 10 days of September: more than 5,000 cuts across Uber, PayPal, Apple, Zomato, and Oracle.

77%

The interesting part is not the spike. It is the shape. When three-quarters of a month's supply comes from one sub-sector, the addressable candidate pool starts to look less like a market and more like a single company graph. That changes how you source it.

Why the pool is not what "laid off" usually means

The May 2026 Meta cohort is a reallocation, not a cost cut, and the September wave is a mix of reallocation runway ending and a rumored second-half round. Treat them as two different cohorts or you will misread the resumes.

Meta cut about 8,000 people on May 20, 2026, roughly 10% of global headcount, hitting Reality Labs, Facebook, recruiting, sales, and non-AI product teams. Revenue is up 22% year over year at $201 billion. Free cash flow is $43.6 billion. Nothing in the financials required layoffs. The AI capex cadence did. That means the laid-off pool is not "weak company." It is "wrong org for where the AI money is going."

Meta also redesigned its internal review structure into four tiers before the reset:

• Top 20%
• Middle 70%
• Lower 7%
• Bottom 3%

The bottom two tiers roughly match the 10% workforce reduction planned across the year. So the cohort you source depends on the month:

The single most important framing for the next 30 days: laid-off does not equal low-performer if you target the May reallocation cohort instead of the September performance cohort. Sourcers who lump them together will bounce off the good ones and load up on the wrong ones.

The severance clock explains the September reply rates

The September spike is partly cohort behavior, not new cuts. Meta's US severance is 16 weeks plus two weeks per year of tenure. May 20 plus 16 weeks lands around September 9. People whose runway is ending are re-activating.

That has direct consequences for outbound timing:

1. Week 2 of severance: reply rates are low. People are decompressing, taking the trip, ignoring LinkedIn.
2. Week 8 to 10: replies pick up but the candidate is still choosy.
3. Week 12 to 14: this is your window. The runway is visibly ending and the pipeline is not full.
4. Week 16 plus: the strong candidates are already in final rounds elsewhere.

If you started outbound to the May cohort in June, most of your no-replies are worth a second touch right now. If you are starting cold in September, lead with the specific team fit, not the "saw you were impacted" opener that everyone else is running.

Where the Meta pool actually lives

In Refolk's index of professional profiles, US senior and staff SWE plus ML engineers with Meta or Facebook in their headline concentrate in five cities, and 79% of that slice sits at one employer. Nationwide outbound is wasted spend.

Here is the index snapshot alongside the public trackers, all pulled for this piece:

The two 79% and 77% numbers landing that close is not a coincidence. Both are telling you the same thing: when Major Internet contracts, the supply-side graph collapses to a handful of employers and a handful of ZIP codes. A recruiter working Menlo Park, Bellevue, Seattle, and NYC captures most of the addressable September pool. Everyone else is paying for reach they cannot use.

79%

This is where alumni-network sourcing outperforms Boolean. When the pool is effectively one company graph, the winning move is to work referral trees: ex-Meta Slacks, Blind threads, TBPN group chats, the bootcamp and PhD cohorts each hiring manager came out of. The people who left in May know who they want to work with in September.

Describing the pool in plain English, and getting back a ranked shortlist across GitHub, LinkedIn, and the open web, is the exact gap Refolk closes. You do not need to build a 40-line Boolean to find the May Reality Labs cohort in Fremont.

Where the demand went, since it is not in the feed

Zero announced hires in the AI Career Hub feed for September does not mean demand is absent. It means Big Tech is hiring quietly through internal transfers and stealth AI orgs, and the Series B and C labs are not posting on trackers at all.

The named examples worth knowing about:

• Meta Superintelligence Labs under Alexandr Wang: the internal destination absorbing headcount while other Meta orgs shed it. The money did not leave, it moved.
• Google's petition of 4,500+ workers asking for layoff protections, including buyouts before mandatory layoffs and guaranteed severance: proof that even non-cut Major Internet talent is mentally on-market. Warm outbound, no severance clock, no stigma.
• KORE1's IT staffing desk (Mike Carter): publicly claims roughly 30 ex-Meta placements since January out of Menlo Park, Seattle, NYC, Austin, with a handful from Bellevue and London. Concrete proof the geographic playbook works.
• Oracle: 30,000 impacted in 2026, the biggest single event. Different sub-sector, different sourcing profile, do not blend it into the Major Internet pool.

Zero announced hires in the tracker is not silence. It is an arbitrage.

The contrarian read on "zero announced hires" is that your candidate pitch this month is not to a job req. It is to a founder at a stealth AI lab who has not written the req yet. Ex-Meta Reality Labs staff engineers who spent three years on hardware-adjacent CV problems are exactly the profile a Series B robotics lab wants, and that lab is not on Layoffs.fyi or LinkedIn.

A 30-day plan that matches the shape of the wave

Do these five things in the next four weeks, in this order, and you will out-source teams that are still keyword-hunting on LinkedIn.

1. Segment the pool by cohort month. Tag the May 2026 reallocation cohort separately from the September performance cohort. Different messaging, different bar.
2. Constrain to five cities. Menlo Park, Fremont, Bellevue or Seattle, NYC, and either SF or Austin depending on your clients. Kill national campaigns for this pool.
3. Time outbound to weeks 12 to 14 of severance. For the May cohort, that is right now. For the September cohort, that lands in early December, and you should already have the list built.
4. Work referral trees, not Boolean. Ex-Meta Slack groups, Blind, Reality Labs alumni chats, the specific PhD cohorts of the hiring managers. Two warm intros beat 200 InMails.
5. Pitch to stealth demand. Series B and C AI labs are the buyers this month. Take the shortlist to founders, not job boards.

Why this window closes fast

The window closes because Q4 hiring at the Series B and C AI labs absorbs the pool, and because the December performance cohort is a noisier signal than the May reallocation cohort. Move now or pay a premium in Q1.

Three specific dynamics compress the window:

• The May cohort is already interviewing. The strong ones have 3 to 5 processes going. Every week you wait, the top decile signs somewhere.
• The September rumored round is performance-weighted. Once that lands, the average signal quality in the pool drops, and screening cost goes up. Your best selection is the pool that exists today.
• Trackers are lagging demand. Because 49% of 2026's layoffs cite AI as a driver (173,465 workers across 188 companies, per Skillsyncer), the reallocation is structural, not cyclical. The employers doing the absorbing will keep hiring quietly through Q1, but they will do it from whoever built the shortlist first.

The 199% spike is not the story. The 77% concentration is not the story. The story is that both numbers tell you the pool is small, clustered, and time-boxed, and the sourcers who treat it that way in the next 30 days will place candidates the teams running national outbound never see.

FAQ

How is this different from the May 2026 Meta cut everyone already sourced?

The May cohort's severance is ending around September 9, which means the people who ignored your June outreach are actively looking now. It is the same list, but reply rates are materially higher than they were in early summer. If your CRM has ex-Meta May cohort no-replies from June, re-touch them this week before someone else does.

Should I bother with the September performance cohort at all?

Yes, but screen harder and ask directly for the internal review tier if you can. Meta's redesigned structure created a lower 7% and bottom 3% tier that maps to the second-half round. The middle 70% is where the strong candidates in the September cohort will sit, and they are worth the work. The bottom two tiers are not, unless you specifically place at hyper-growth startups where velocity outweighs prior review scores.

What about Google, given the petition?

The 4,500+ signature petition is a signal that Google talent is mentally on-market without being on the layoff feed, which makes them warm outbound with no severance clock. Reply rates are lower than the ex-Meta cohort right now because they still have jobs, but the ceiling is higher and there is no stigma to manage. Run them as a parallel list, not the primary one.

Does the Major Internet concentration hold for non-engineering roles?

Partly. The May Meta cut hit recruiting, sales, non-AI product, integrity, cybersecurity, and content design alongside Reality Labs. Recruiting and integrity roles do not have the same five-city concentration as senior SWE and ML, so the geographic playbook loosens. For non-engineering, work the alumni networks harder and rely less on city filters.