The Cycle That Completed Faster Than Anyone Expected
The tech layoff wave of 2022-2023 was genuinely large. Meta cut 21,000 employees. Amazon cut 27,000. Google cut 12,000. Microsoft cut 10,000. The numbers were staggering, and the human cost was real. By mid-2024, the same companies — and many that cut even more aggressively — are posting hiring plans that partially reverse those reductions. The cycle completed in approximately 18 months. The proximate cause is AI. Every major tech company has concluded that AI development requires more engineering headcount than they currently have — specifically, ML engineers, AI researchers, and the infrastructure engineers who can build and maintain large-scale model training and inference systems. The cuts of 2022-2023 were real-estate for AI investment.
Traditional software engineering roles — CRUD applications, frontend work, basic backend services — face genuine pressure from AI-assisted development tools. An engineer who was writing routine code three years ago is in a different market than they were. AI-adjacent roles are in a hiring frenzy. ML engineering, AI safety research, prompt engineering, and AI product management are all commanding compensation that would have seemed absurd in 2020. This is creating a bifurcated engineering market: increasing difficulty for engineers whose work is most automatable, increasing demand and compensation for engineers whose work requires the combination of technical depth and domain judgment that AI tools cannot yet replicate.
What This Means for Career Strategy
The lesson is not that tech is safe or unsafe. It is that tech is, as always, in the process of redefining itself, and the engineers who move toward what is being built rather than away from what is being automated will be fine. The specific skills that are increasing in value: systems understanding that allows working with AI tools rather than being replaced by them, the ability to evaluate AI output quality in specific domains, and the combination of software engineering depth with machine learning literacy that allows building AI-integrated products rather than simply AI-generated code. These skills are learnable, and the engineers who invest in learning them over the next two years will be better positioned than those who treat the AI tooling wave as something to wait out.