AI is supposed to replace developers. But the data says otherwise.

We’ve heard that AI will automate away engineering roles. But according to SignalFire’s analysis of millions of employees at over 80 million companies, engineering is one of the strongest performing sectors in tech — even outperforming tech as a whole.

1The numbers

Tech employment dropped 25% from 2019 levels. Meanwhile, engineering roles only decreased by 11%. And the trend continues up the stack: at the 12 largest tech companies (Alphabet, Meta, Apple, Amazon, Microsoft), engineers accounted for 55% of all hires in 2025, up from 46% in 2019. Even at early stage startups, engineering hiring increased 7% from 2019. “We’re seeing some things on the ground that contradict that storyline,” said Asher Bantock, SignalFire’s Head of Research.

The trend also appears in other metrics. According to Nvidia CEO Jensen Huang, “our software engineers are busier than ever” after deploying agentic AI, which was supposed to automate their work but has actually broadened the range of projects they can tackle. Anthropic’s Head of Economics, Peter McCrory, looked at unemployment rates in occupations heavily impacted by AI versus those with low exposure and couldn’t find any statistically significant difference.

2Why the math works out this way

This phenomenon is known as the Jevons Paradox, which says that improvements in efficiency frequently lead to increased demand rather than reduced demand. If AI makes individual engineers more productive, then there is greater demand for engineers overall. This can manifest in two ways: companies hire more engineers, or they pursue larger, previously unfeasible projects because they have AI leverage.

This reminds us of the introduction of spreadsheets in the 1980s. Financial analysts worried that automating accounting would cause layoffs, but the technology actually made financial modeling accessible to many more companies, increasing overall demand for financial workers.

3What this means if you build things

So don’t panic — AI won’t automate engineering roles! Well, maybe not. What we can say is that successful engineers will be those who learn how to use AI to amplify their impact, shipping more code, solving harder problems, and becoming more valuable in the process. The SignalFire numbers reflect net employment changes, not absolute numbers, so we can’t tell whether there’s a growing divide within the engineering community between those who have adopted AI tools and those who have not. But there definitely is such a divide, and it’s likely getting bigger.