February 19, 2026
Robots code, bottlenecks win
CTO Says 93% of Developers Use AI, but Productivity Is Still 10%
93% use AI, still only 10% faster—devs blame bottlenecks and burnout
TLDR: A new study says almost all developers use AI and over a quarter of production code is machine‑written, yet overall speed is up only about 10%. Commenters say AI isn’t the problem—bad specs, slow processes, and low morale are—sparking debate over whether AI accelerates work or just exposes organizational flaws.
CTO Laura Tacho dropped a bomb at Pragmatic Summit: 93% of developers use AI and 26.9% of production code is AI‑written, yet productivity has stalled at about 10%. Her research of 121,000 devs says time saved is roughly 4 hours a week, onboarding time is cut in half, and AI works best in well‑run companies—while messy orgs just get their mess exposed. But the real fireworks are in the comments.
The crowd’s vibe? “AI isn’t the bottleneck—your org is.” User overgard nailed it with the bottleneck mic drop, while gedy reminded everyone that typing code isn’t the hard part—figuring out what to build is. nasretdinov warned AI is the new “time‑saver that bites later,” like those “fast and loose” tools that age badly. And downrightmike brought the heat: if management keeps saying AI can replace devs, why would anyone feel motivated?
Some readers even called the headline misleading, with bluejekyll clarifying the point: it’s not that AI “doesn’t work,” it’s that the gain is just 10%. Meanwhile, jokes flew about AI being a tireless junior who “writes faster—just not the right thing,” and that no model on Earth can cancel a meeting or fix bad specs. Consensus: AI helps, but culture, process, and clarity still rule.
Key Points
- •Research surveyed 121,000 developers across 450+ companies; 92.6% use AI coding assistants monthly and ~75% weekly.
- •Developers report saving about 4 hours per week; productivity gains rose ~10% early in AI adoption and have remained around that level.
- •AI-authored code reached 26.9% of production code between Nov 2025–Feb 2026, up from 22% in the prior quarter.
- •Daily AI users have nearly one-third of merged, production-bound code written by AI.
- •Onboarding time (measured as time to 10th PR) has been halved from Q1 2024 to Q4 2025; organizational impacts vary widely depending on AI adoption quality, affecting incident rates.