SignalFire's 12 Tech Majors Now Hire 55% Engineers. Stop Mining WARN.

If your sourcing plan for Q3 is "scrape every WARN notice and ping the names," the SignalFire State of Talent 2026 report, covered by TechCrunch on June 24, just invalidated it. Engineers are the most resilient corporate function inside the twelve biggest tech employers in the United States, not the most exposed. The densest passive engineering pool in 2026 sits where you stopped looking: on the org charts of Alphabet, Meta, Apple, Amazon, Microsoft, Netflix, Nvidia, Tesla, Uber, Airbnb, Block, and Stripe.

What the June 24 data actually says

SignalFire's Head of Research, Asher Bantock, ran careers data across 80 million companies. The headline number from the report, which TechCrunch led with: engineers were 55% of all new hires at the 12 Tech Majors in 2025, up from 46% in 2019. Total hiring at those companies fell 25% over the same window. Engineering hiring fell only 11%.

That gap is the whole story. Every other function (product, design, marketing, sales operations, recruiting itself) absorbed the cuts. Engineers absorbed almost none of them. Early-stage startups actually hired 7% more engineers in 2025 than in 2019, while design dropped 22% and marketing dropped 18% over the same period.

55%

The implication for sourcing is direct. The signal you want, "engineer who recently joined or is currently working at a Tech Major," is more concentrated in 2026 than at any point in the last seven years. The signal "engineer recently let go from a Tech Major" is rarer than the headlines suggest. At Block's most recent layoff round, engineers were less than 30% of the cuts despite being a much larger share of headcount. The WARN-tracker pool is over-indexed on PMs, designers, marketers, and middle managers. If you're sourcing senior backend talent from that list, you're getting adverse selection.

The AI-layoff narrative is mispricing your pipeline

You will get pushback on this in a hiring committee. Someone will mention Oracle's 10-K, which logged 21,000 cuts citing AI. Someone else will note Meta eliminated 8,000 jobs this month, with software engineers explicitly named. Both are true. Neither contradicts the SignalFire base rate.

The headline cuts are loud because they are framed as AI-driven. The underlying engineering hiring continues. Nvidia's Jensen Huang put it crisply at Stanford GSB in April: with agentic AI in the loop, "software engineers are busier than ever." Anthropic's own Head of Economics, Peter McCrory, told TechCrunch in March he sees no significant AI-driven workforce effects, which is awkwardly at odds with Dario Amodei's earlier warning about 50% of entry-level white-collar jobs. When the company selling the disruption disagrees with itself about whether the disruption is happening, that is your cue to trust the careers data over the press release.

The Jevons frame matters here. AI coding tools raise the floor on what one engineer is expected to ship. That makes each engineer harder to justify cutting, not easier. The 55% number is the downstream effect.

The cuts are happening in PM, design, and middle management. The hires are still engineers. Source accordingly. </pull> Wait, that block should be a pull, not a quote tag. Let me restate the point in the prose: the cuts hit coordination roles, the hires stay technical, and your sourcing strategy should be weighted toward the second fact.

pull The WARN list is over-indexed on PMs and designers. The hiring is still engineers. Source where the hires are.

## The persona has shifted under the same job title

Here is where most sourcing operations are still running 2021 playbooks against a 2026 reality. Inside the engineering 55%, the role mix has been redrawn.

Since ChatGPT shipped in late 2022, the share of AI/ML Engineer roles is up 39% across SignalFire's tracked population. Research Engineer is up 28%. Forward-Deployed Engineer is up 30%. Sales Engineer is up 11%. Front-end Engineer is down 25%.

Read that last number again. A quarter of the front-end pool inside these companies, gone from the new-hire mix. If your Boolean string still leads with "React engineer at Meta" or "Vue at Stripe," you are searching a sub-pool that has been shrinking for three years. The growth is in titles that barely existed at volume in 2022.

### The "Super IC" is the modal target

Entry-level hiring at the Tech Majors is down roughly 65% since 2019. New-grad share of new hires fell from about 11% in 2022 to about 7% in 2024. Each engineering manager at a Tech Major now spans about 12 reports, up from 10. At startups, it is closer to 15. The flat-org effect is real.

What this means for your persona doc: the modal engineer being hired at one of these 12 companies is a senior IC who has absorbed work that used to belong to a junior engineer, a PM, and sometimes a designer. SignalFire's Heather Doshay calls this the "Super IC" pattern. Sourcing against it requires queries that look more like "Staff engineer with 10 plus years who has shipped a product surface end to end and references a PRD they wrote themselves" than "senior engineer, FAANG, 5 plus years."

This is exactly the friction we built [Refolk](/) to remove. You describe the person in plain English, including the soft signals (owned a roadmap, shipped without a PM, presented at an internal eng all-hands), and you get a ranked shortlist across GitHub, LinkedIn, and the open web. The Boolean translation step that used to lose 80% of the nuance happens inside the model, not inside your head at 11pm.

Why the 12 are the right passive pool, with one honest caveat

Standard recruiting research puts 70% of the workforce in the passive bucket. Inside the Tech Majors, engineers are stickier than that average. Attrition runs around 9% for engineers versus around 13% for sales and design at the same companies. Engineers are, in SignalFire's phrasing, "hired faster and turning over more slowly than any other corporate function."

The honest read of that number cuts both ways.

The good news: when you do close a Tech Major engineer, your expected retention is meaningfully higher than the base rate. You are sourcing from a pool that has already self-selected for staying put.

The harder news: response rates will be lower than you are used to. LinkedIn InMail benchmarks sit between 10 and 25% for genuinely personalized messages, and 5.1% for cold email. Against an engineer with 9% attrition and four recruiter pings already in their inbox this week, expect the floor of that range, not the ceiling. The math still works because the quality is higher, but plan your outreach volume accordingly.

9%

Where to actually find the 207,000

Internal professional-network data shows roughly 207,000 Senior and Staff Software Engineers currently working in the US who match senior IC patterns, with top employers including Google, Datadog, and the rest of the named Tech Majors. That is your ground-truth pool size for sourcing engineers from big tech. The job is not finding more names. The job is ranking the names you can already pull, against a persona that is more specific than "senior engineer at Google."

Three concrete moves for the next sprint:

1. Re-weight your sourcing dashboard so the 12 Tech Majors are over-represented relative to layoff-tracker companies. If your current pipeline is 60% ex-laid-off candidates, that ratio is upside down given the 55% hiring concentration.

2. Refresh your role taxonomy. Add Forward-Deployed Engineer, Research Engineer, and AI/ML Engineer as first-class targets. Demote front-end-only searches unless the role specifically requires it.

3. Rewrite the first message. The Super IC has absorbed PM and design work. Lead with the scope of ownership in your role, not the compensation band. Engineers with 9% attrition do not move for a 10% bump. They move for scope they cannot get internally.

Refolk fits move two cleanly: when the role taxonomy is shifting faster than your ATS tags can keep up, plain-English search lets you query "people doing forward-deployed work even if their title still says Solutions Engineer" without rebuilding your Boolean library every quarter.

The counter-evidence you should not ignore

A responsible read of the SignalFire data acknowledges three things.

First, the entry-level collapse is real. If your req is for a new-grad or a one-to-three-year engineer, the Tech Major pool is the worst place to source it. The 65% drop in entry-level hiring since 2019 means those engineers either never got hired or already left for startups. Tawni Cranz, SignalFire Operating Partner and former Netflix CHRO, has been explicit that new-grad portfolios now matter more than pedigree for those roles.

Second, the role-mix shift cuts against some specialties. A front-end specialist with no AI/ML exposure is in a shrinking sub-pool inside these companies. The aggregate "engineering is resilient" headline does not protect every individual.

Third, the cuts that did happen were not random. When Meta moved 8,000 people out this month, software engineers were specifically named alongside managers. The 55% hiring number and the named-cut headlines are both true simultaneously. The companies are reshuffling, not just growing.

The sourcing implication holds anyway. A reshuffle creates motion in the pool. Motion is what passive candidate sourcing converts on. The WARN-tracker reflex catches the wrong half of that motion. The 12-company watchlist, refreshed weekly, catches the right half.

What to do Monday

Pull your last 90 days of outbound. Count what percentage of accepted intros came from candidates currently employed at one of the 12 Tech Majors versus candidates recently laid off. If the ratio is not roughly tracking the 55% hiring concentration, your top of funnel is mispriced.

Then rewrite one persona this week using the Super IC frame. Run it against your sourcing tool of choice. If you are using Refolk, the query is one sentence. If you are not, budget an afternoon for the Boolean. Either way, the input that matters is the persona, not the keyword list.

The AI-replaces-coders narrative will keep generating headlines through 2026. The 55% number will keep being true underneath it. Source the mix, not the cuts.

FAQ

Who are SignalFire's 12 Tech Majors?

Alphabet, Meta, Apple, Amazon, Microsoft, Netflix, Nvidia, Tesla, Uber, Airbnb, Block, and Stripe. SignalFire treats them as a coherent benchmark pool for tracking talent flows because together they employ a meaningful share of senior US software engineers and their hiring patterns tend to lead the broader industry by one to two quarters.

Does the 55% figure mean these companies are growing engineering headcount?

Not necessarily. Total hiring at the 12 fell 25% versus 2019, and engineering hiring still fell 11%. The 55% is a share of the (smaller) new-hire pool, not absolute growth. The point for sourcers is that engineers are over-represented in what hiring is still happening, and under-represented in the cuts. Your candidate pool concentration matches that mix even when net headcount is flat or declining.

How do I source engineers from the Tech Majors without burning my LinkedIn quota?

Three tactics. One, query for current employees against a specific persona (AI/ML, Forward-Deployed, Staff IC who shipped a named product) rather than scraping by company. Two, prioritize candidates with public artifacts (GitHub, talks, papers) so your first message can reference real work. Three, lower your outreach volume but raise the per-message investment. Engineers with 9% attrition do not respond to template mail. Tools like Refolk are designed to do the persona-to-shortlist step in one query, which is where most of the time savings sit.

What about Oracle's 21,000 cuts and Meta's 8,000 this month?

Both are real and both are partly engineering. They do not invalidate the SignalFire base rate because, at the same companies and across the other ten, engineering hiring is still the largest single bucket. The honest read is that the industry is reshuffling engineers, not eliminating them. For sourcing, that means motion in the pool, which is the precondition for any outbound strategy to convert. Watch the named cuts for short windows of availability, but build your steady-state pipeline against the 55%.