The outside world has been debating whether AI has begun to replace jobs, and the answer given by the latest hiring data does not point to "engineers being replaced" as many expected. Although the number of layoffs in the technology industry hit a multi-year single-month high in May, and "AI" has become one of the most frequently mentioned reasons for layoffs, judging from recruitment trends, engineering positions will be the most resilient function in the technology industry in 2025.
SignalFire researchers, who tracked the career trajectories of millions of employees at more than 80 million companies, found that overall hiring at tech giants fell 25% compared with 2019, but hiring for engineering roles fell only 11%. This means that, in an environment of overall contraction, engineering jobs are far more resilient to stress than most other functions. The reason why SignalFire pays more attention to hiring data is because status updates after layoffs often lag, and looking at layoffs alone cannot truly reflect the immediate changes in the labor market.
The report also shows that among the 12 leading technology companies defined by SignalFire, engineers will account for 55% of new employees in 2025, up from 46% in 2019. The 12 companies include Alphabet, Meta, Apple, Amazon, Microsoft, Netflix, Nvidia, Tesla, Uber, Airbnb, Block and Stripe. Among early-stage startups, demand for engineers is not shrinking, with the number of engineers they will hire increasing by 7% in 2025 compared with 2019.
Asher Bantock, head of research at SignalFire, believes that if AI will really replace engineering talents on a large scale, engineering jobs should be the first to see a significant decline amid the current overall shrinkage in the technology industry, but the actual data does not support this judgment. He said that outsiders often say that the reason for layoffs is AI, especially "AI can replace code work," but the situation he sees on the front line is inconsistent.
This view is also in contrast to the public statements of some AI company executives. Anthropic CEO Dario Amodei has warned in the past that AI could eliminate half of all entry-level white-collar jobs within five years and push unemployment to 20%. But Peter McCrory, chief economist at Anthropic, said he has not yet observed a significant impact on the labor market from AI. He added at the time that there was no significant difference in unemployment rates between technical writers, data entry clerks and software engineers who use Claude to automate their core tasks, and manual jobs with less exposure to AI.
Nvidia CEO Jensen Huang’s attitude is more direct. In an interview with Stanford Business School in April this year, he clearly denied the statement that "AI will destroy software engineering jobs" and said that Nvidia engineers are already using intelligent agent tools, and as a result, they are "busier than ever." Huang Renxun also said that although AI agents can generate code quickly, they also constantly force engineers to come up with "the next idea."
The current software engineering industry is very much like a real-life example of Jevons' paradox: efficiency improvements have not reduced demand, but have continued to expand workloads. In other words, AI tools make engineers more efficient, but also allow them to take over more work, ultimately resulting in "more efficiency means more work." This is why, at a time when AI is at its hottest, engineering jobs have not shrunk as many people expected, but have become the most impact-resistant jobs in the technology industry.