State of AI Impact in Engineering: Q2 2026 Report
DX's State of AI Impact in Engineering: Q2 2026 Report (July 2026; n=500+ engineering organizations, telemetry plus surveys) found that AI-generated code reached 52.7% of all code while median quarterly AI spend grew nearly 28x in a year in the tech sector and the share of time on new feature work stayed at roughly 57–58%.
~2.2
1.2
Key findings
- 01AI-generated code accounted for 52.7% of all code, up from 24% two quarters earlier.AI utilization is described as above 90% across the industry.
- 02Median quarterly AI spend grew from roughly $1.5K to $44K in a year in the tech sector, a nearly 28x increase.Over the same period the innovation ratio (new feature work vs. maintenance) held at roughly 57–58%.
- 03Median weekly throughput rose 37%, from 1.42 to 1.94 pull requests per engineer per week.Heavy AI users save over 6 hours a week, but DX says saved hours are not yet visibly turning into new value.
- 04Organizations with 15–99 engineers hit about 2.2 PRs per engineer per week, versus 1.2 for those with 750+ engineers.Smaller organizations paid more per AI seat but got more throughput per dollar.
- 05Change confidence fell 6.1% while code maintainability improved 3.8%; the Developer Experience Index fell from 67 to 65.Declines were driven by incremental delivery, local iteration speed, and review turnaround.
By the numbers
~28xgrowth in median quarterly AI spend (tech sector, one year)
37%rise in weekly PRs per engineer
−6.1%change in developers' confidence their changes won't break things
What it means for you Draft
For executives at $10–100M companies
Output and AI spend are rising fast in this data, but the share of time going to new features has not moved, so the payoff is still unclear. For a $10–100M company with a small engineering team, the more encouraging signal is that smaller engineering organizations were getting more throughput per AI dollar than large ones. Pairing any AI spend with a quality and delivery measure seems prudent.
For practitioners
Watch review turnaround, PR size, and change failure rate as AI-written code grows. Track spend per engineer against throughput so tool costs don't outrun results.
Limitations
Medium trust.Large multi-organization telemetry plus surveys, but customer-only sample, gated methods, and publisher sells AI-impact measurement software.
Data comes from DX customers, not a random sample, and the full report is gated. Much of the data is correlational, and figures such as time saved are partly survey-based.
About the publisher: Sells developer-productivity and AI-impact measurement software; data comes from its customers.
DX. "State of AI Impact in Engineering: Q2 2026 Report." July 22, 2026. https://getdx.com/report/state-of-ai-impact-in-engineering-q2-report/
Related reports
Work AI Index 2026: Botsitting, Botshitting & the Hidden Human Labor of AI at WorkWorkers report 11 hours a week saved but 6.4 hours spent supervising and fixing AI; only 13% see significant organizational gains.
The AI Productivity Paradox Report 2025High-AI teams merged 98% more pull requests, but review time rose 91% and company-level delivery metrics did not improve.
Technology Report 2026Leaders expect a 95% developer productivity uplift in 1–2 years; current gains are 20–27%. Bain says AI moves bottlenecks into review and coordination.