New hiring data from major tech companies dismantles the “AI kills coding jobs” narrative — and points to a deeper structural shift in what engineers actually do.
JUNE 2026 — TECH HIRING DATA
−25%
Overall tech hiring vs. 2019
−11%
Engineering roles decline (vs. 2019)
55%
Engineers as share of all 2025 hires
41%
Middle manager roles cut
What the Data Actually Shows
New hiring data from major tech companies, reported by TechCrunch on June 24, 2026, reveals a striking paradox: overall tech headcount is down 25% from 2019 levels — but engineering roles have fallen only 11% over the same period. More tellingly, engineers comprised 55% of all new hires across large tech firms in 2025.
The timing matters. This data lands directly against McKinsey’s Skill Change Index, which flagged SQL and software development as having the highest automation exposure in the workforce — above 30%. If AI was supposed to hollow out engineering, the hiring departments didn’t get the memo.
The real displacement is happening one layer up. Middle management — the layer that translated business strategy into engineering roadmaps — is being cut at 41%. That gap isn’t going unfilled. It’s being absorbed by engineers themselves.
The key insight: AI automates coding tasks but amplifies engineering judgment — the companies cutting headcount are cutting the roles that no longer require human reasoning, not the ones that do.
The Structural Read
The paradox resolves when you distinguish between coding and engineering judgment. Cursor and GitHub Copilot write the boilerplate. Block’s BuilderBot generates roughly 15% of Block’s production code. None of that triggered layoffs at Block — because what engineers actually do is harder to automate than the code they write.
Engineers frame the problem. They translate ambiguous product intent into precise system constraints. They recognize when a working build is still architecturally wrong. They make the tradeoffs that don’t fit into a prompt. AI makes each of these tasks faster — it does not make them unnecessary.
The layer being eliminated is the one that stood between business intention and engineering execution: middle managers who ran sprint planning, wrote PRDs, and held weekly syncs to align “the business” with “the tech team.” With AI collapsing that translation gap, companies find they need fewer translators and more engineers who can work directly from intent to implementation.
ROLE RESILIENCE — DECLINE VS. 2019
Three Implications That Follow
ENGINEERS GAIN STRATEGIC SCOPE
As management layers thin, engineers inherit product decisions, customer context, and business tradeoffs. The old separation between “technical” and “business” roles is collapsing. The engineers who survive — and thrive — are those who can hold both.
AI TOOLS RAISE THE FLOOR, NOT THE CEILING
Cursor and similar tools compress the time from spec to working code. But the ceiling — the quality of the spec, the soundness of the architecture — still requires a skilled engineer to set. AI raises average output per engineer, making each hire more productive and harder to justify cutting.
THE MCKINSEY INDEX MEASURES THE WRONG THING
Automation exposure metrics measure whether AI can perform task components — they don’t measure whether companies will stop hiring the humans who perform those tasks. Engineering’s “high automation exposure” applies to syntax and boilerplate, not to architectural judgment. The market is already pricing this difference in.
Loops Paradigm
From Roadmaps to Loops
The classic product cycle — discovery, spec, build, ship, measure — assumed a long handoff between roles. With AI, the loop compresses: engineers ship to production faster, measure in real time, and iterate within hours. This is the loops paradigm in action. It doesn’t need fewer engineers — it needs engineers who can run the full loop.
Business Engineer Framework
The Builder-PM Manifesto
This data is a live case study for the Builder-PM thesis: the engineer who can frame a problem, ship a solution, and read the outcome is the most defensible role in tech. Specs become working branches. Roadmaps become loops. The human who frames the work becomes more valuable, not less.
The Bottom Line
The “AI kills engineering jobs” narrative was always a misread of the automation exposure data. What AI kills is undifferentiated task execution — the kind that middle management layers were largely built to coordinate. The engineer who can reason about systems, hold product context, and close loops fast is precisely the profile large tech companies are doubling down on. When 55% of your new hires are engineers in a year when you have every AI tool available, the signal is clear: engineering judgment doesn’t automate, it compounds.
Source: TechCrunch, June 24, 2026 — tech hiring data across major companies. McKinsey Skill Change Index via McKinsey Global Institute.