AI Was Meant to Kill Dev Jobs. Data Says Otherwise.

Engineering jobs were supposed to be among the first casualties of AI. Instead, they're holding up better than almost any other category in tech. That's the conclusion of a new SignalFire analysis covering millions of workers across more than 80 million companies — a dataset large enough to cut through the noise of anecdote and headlines.

The numbers

Overall tech hiring fell 25% compared to 2019. Engineering roles fell by just 11% — less than half the sector-wide decline. The gap is sharper at the top: at the 12 largest tech companies, including Alphabet, Meta, Apple, Amazon and Microsoft, engineers made up 55% of new hires in 2025, up from 46% six years ago. At early-stage startups, engineering hiring grew 7% compared to 2019. 'What we're seeing on the ground is a little inconsistent with' the AI-replacing-engineers narrative, said Asher Bantock, SignalFire's head of research.

Other data points point the same direction. Nvidia CEO Jensen Huang has said that software engineers at his company are 'busier than ever' since implementing agentic AI — the tools that were supposed to replace them are instead expanding the scope of what gets built. Peter McCrory, Anthropic's head of economics, compared employment outcomes in roles heavily exposed to AI against those in less-exposed positions and found no statistically significant unemployment effect.

Why the math works out this way

Economists have a name for this dynamic: the Jevons paradox. The principle is that efficiency gains often increase total demand rather than reduce it. When each engineer can do more with AI assistance, the value of having an engineer goes up, not down. Companies respond by hiring more engineers, or by taking on more ambitious projects they wouldn't have attempted without AI leverage.

The pattern echoes what happened when spreadsheet software arrived in the 1980s. Finance professionals feared automation would eliminate accounting jobs; instead, the software made financial analysis cheap enough that far more businesses could afford it. Total headcount in finance grew. The automation created demand that hadn't existed before.

What this means if you build things

The practical reading here isn't 'relax, AI won't touch engineering.' It's more specific: the engineers who are thriving are those who learned to use AI as a force multiplier — they ship more, take on bigger problems, and become more valuable as a result. The SignalFire data doesn't measure all engineers equally; it measures net hiring. What it can't see is the internal split between engineers who have integrated AI tools into their workflow and those who haven't. That split is real and probably widening.