AI tools writing code faster won't replace software engineers anytime soon, according to Bernstein analysts. They argue coding accounts for only about 20% of an engineer's typical workload, with the majority spent on documentation, deployment, integration, and process management. Even with AI accelerating coding tasks, it would automate only a fraction of the job.

Surveys indicate nearly 80% of developers already use AI tools, but most report generated code requires significant refinement and debugging. While tech layoffs remain elevated since the post-COVID hiring surge, Bernstein notes much of the reduction reflects excess hiring and weaker demand, not wholesale AI-driven replacement. Sectors linked to AI have seen higher recent cuts, but tools aren't replacing teams at scale yet.

The firm also rejected comparisons to the dot-com bubble, highlighting leading AI companies are established, profitable entities with strong cash positions, not speculative startups. Bernstein expects AI to become an incremental layer within enterprise technology rather than an immediate substitute for human knowledge work.