Harvard study of 718 firms: AI produces more code but does not speed up delivery

Oct 19, 2026 · Issue 12 · 12 sources, 12 articles

A Harvard study based on Jellyfish data, covering 718 firms, found that AI assistants and agents increase the volume of code, but downstream bottlenecks mean this does not deliver productivity gains. Further analyses say coding accounts for 10–20% of delivery time, and technical debt grows alongside AI-accelerated code.

Who is affected and how

CTO

Measure the return on AI investment across the whole delivery process (review, testing, release), not by the volume of generated code.

tech lead

Review and testing capacity must be expanded, because faster code generation causes congestion there.

QA tester

With the increased volume of code, test automation and risk-based testing gain priority.

Narrative: More code, but not more delivered software

Sources

• Study Finds AI-generated Code Fails to Boost Productivity Due to Downstream Bottlenecks - InfoQinfoq.com · Oct 22, 2026
• Study Finds AI-Generated Code Fails to Boost Productivity Due to Downstream Bottlenecks | LavX Newsnews.lavx.hu · Oct 22, 2026
• AI Coding Productivity Study: 30% More Code, Flat Outputayautomate.com · Oct 25, 2026
• AI Can Write Code Faster. So Why Isn't Software Shipping Faster?hackernoon.com · Oct 19, 2026
• AI Can Write More Code. Why Aren't Teams Shipping More Software? - FloNeofloneo.co · Oct 19, 2026
• AI in Software Development: 87% Code Faster, 72% Face More Frictiontheoutpost.ai · Oct 22, 2026
• AI in Software Engineering: The Reliability Paradox | Bain & Companybain.com · Oct 19, 2026
• AI in Software Engineering: The Reliability Paradoxbain & company · Oct 19, 2026

Machine-generated summary based on the titles and excerpts of the source articles. The facts come from the linked article.