A new analysis from economists converts stock market movements into an estimate of how much AI will improve software engineering productivity. The figure: a 32.6% boost, as priced in by investors. The calculation treats share price changes as a proxy for expected gains, rather than measuring actual output changes.
The approach is clever but fragile. Markets can get carried away, the economists note, and stock prices reflect many factors beyond AI. The 32.6% is an inference from investor behavior, not a forecast from engineering benchmarks.
For hardware and compute, the number matters because it implies sustained demand for AI infrastructure. If investors truly expect a third more output per engineer, they are betting on heavy deployment of AI tools and the chips that power them. But the caution about market overreaction applies equally to hardware spending plans.