Plot twist: Engineering Is the Most Resilient Job in Tech. The Layoff Headlines Had it Backwards.

Ronni Holmvig Strøm · 2026-06-26

In May, tech layoffs hit their highest single-month total in years, and AI was the most-cited reason, according to outplacement firm Challenger, Gray & Christmas. At first

In May, tech layoffs hit their highest single-month total in years, and AI was the most-cited reason, according to outplacement firm Challenger, Gray & Christmas. At first glance, the plot seemed straight forward. Coding tools got good, one engineer now does the work of several, the rest are surplus. Then SignalFire went and looked at who is actually getting hired. Engineering, it turns out, was the most resilient job function in tech in 2025.

The Hiring Data Says the Opposite of the Press Releases

SignalFire's State of Talent Report tracked the careers of millions of employees across more than 80 million companies. The firm deliberately worked from hiring rather than layoffs, on the reasonable theory that people delay updating their status after a cut, which makes layoff counts a lagging and noisy signal. Hiring is what a company does when it has decided what it needs next.

The numbers lean hard against the substitution thesis. Total hiring across large tech fell 25% from 2019 levels. Engineering roles fell only 11%. Across the twelve firms SignalFire calls Tech Majors, e.g. Alphabet, Meta, Nvidia, Stripe, engineers made up 55% of all new hires in 2025, up from 46% in 2019. At early-stage startups the pull was stronger still: 7% more engineers hired in 2025 than in 2019, during a contraction that thinned almost everything else.

If AI were genuinely substituting for engineering talent, that function would be the first to fall in a downturn, not the last. As SignalFire's head of research Asher Bantock put it, "What we're seeing on the ground is a little inconsistent with that." The companies blaming AI for cuts are, at the same time, hiring engineers as a larger share of new hires than they did before the tools existed.

Cheaper Code Is Not Less Code

This is the Jevons paradox arriving on schedule. When a resource gets cheaper to use, total demand for it tends to rise, because the cheaper work expands to fill the new capacity. Efficient steam engines burned more coal, not less. Productive engineers write more software, not less, and the backlog of things worth building was never the binding constraint anyway.

The doom forecast made a category error. It read a falling cost-per-task as a falling demand for the task. Those move together only if the appetite for software is fixed, and it has never once been fixed. Every prior jump in developer productivity, from compilers to cloud infrastructure, was supposed to reduce the need for engineers. Each one enlarged it, because it lowered the price of attempting things that were previously too expensive to try.

What changes under AI is where the bottleneck sits. Nvidia's Jensen Huang, who has every incentive to talk his book but is describing his own engineering floor, said his engineers are "busier than ever" now that they all run agentic tools. The agents write code close to instantly, which moves the scarce step upstream, to deciding what is worth building. Bantock's line is the cleaner version: engineers are "suddenly a lot more productive, and there's endless work for them to do."

Where the Honest Caveat Lives

The aggregate is healthy. The composition is the open question though. SignalFire's figures cover engineering as a function, not the junior rung specifically, and Anthropic's Dario Amodei has warned that AI could erase half of entry-level white-collar jobs. Worth holding that next to his own head of economics, Peter McCrory, who told TechCrunch in March he had seen no material difference in unemployment between high-exposure roles like software engineering and jobs that need physical dexterity.

For a field that was supposed to be automation's first casualty, engineering is the function being handed a larger canvas.