The data is stark and we shouldn’t look the other way. So far in 2026, more than 92,000 workers in the technology sector have lost their jobs. These aren’t predictions or theoretical models; these are announced and counted layoffs. Snap eliminated 16% of its entire workforce last month, with CEO Evan Spiegel explaining bluntly that advances in AI allow smaller teams to do the same work. Meta cut 8,000 jobs while planning to spend $135 billion on AI infrastructure this year. Amazon has eliminated at least 30,000 roles since October. Microsoft offered early retirement to 7% of its US employees. Oracle restructured 25,000 positions.
What makes this wave of layoffs different from previous ones is the profile of those affected. It’s not about automating manual or repetitive tasks, something economists have been debating for decades. This time AI is replacing white-collar workers: writers, analysts, programmers, designers, support staff. And it’s doing so at a pace that surpasses any previous automation wave. Economists already speak of a «permanent structural transformation», which in more direct terms means those jobs aren’t coming back when the economic cycle improves.
The most uncomfortable contradiction in all of this is that the very companies laying off thousands of people are the ones investing hundreds of billions in AI. The money has to come from somewhere, and the payroll is one of the most controllable costs on any balance sheet. The research presented this week at ICLR 2026 adds another layer of complexity: the most advanced AI models that reason better turn out to be less reliable when using external tools, which raises serious questions about whether companies are deploying these systems with due caution or simply rushing to avoid being left behind.
Article published on 29 April 2026 | Technology blog
