The State of AI Impact in Engineering: Q2 2026

DX's Q2 2026 report on AI impact across 500+ engineering teams finds that over 50% of code is now AI-generated, up from 34% in Q1 2026, but median pull request sizes have nearly doubled and the Developer Experience Index dropped from 67 to 65 over four quarters, indicating quality and trust may be declining despite velocity gains.

The State of AI Impact in Engineering: Q2 2026 Data from 500+ teams reveals that AI is delivering measurable velocity gains, but velocity alone isn't the story. Welcome to the latest issue of Engineering Enablement, a weekly newsletter sharing research and perspectives on developer productivity. 🗓 Join me and Brian Houck on July 23 https://getdx.com/webinar/ai-in-engineering-q2-2026-benchmarks-research-readout/?utm source=newsletter for a readout of this report, where we’ll discuss new findings from DX’s data on AI tool usage, spend, and impact across 500+ organizations. Register here. https://getdx.com/webinar/ai-in-engineering-q2-2026-benchmarks-research-readout/?utm source=newsletter We are excited to announce our Q2 2026 AI impact report. When we first began tracking the impact of AI on engineering teams, our primary goal was to measure AI cohorts against historical baselines to answer the question of what happens to software output after adoption. With industry-wide AI adoption exceeding 90%, comparing AI users against a non-user control group is no longer a viable measurement strategy. Engineering leaders are now under immense pressure to justify exponentially-increasing AI budgets. The data from our Q2 report reveals that while AI is delivering objective gains in velocity, those gains are highly uneven. Download the full analysis here. https://getdx.com/report/state-of-ai-impact-in-engineering-q2-report/?utm source=newsletter In the new report, we’ve uncovered a number of critical trends, including: 1. Over 50% of code is now generated by AI. This metric has accelerated rapidly, increasing from 34% in Q1 2026 to 52% in Q2 2026. This steep trajectory indicates that once AI tools are deployed, the code they generate rapidly scales across codebases, frequently moving through reviews, dependencies, and shared workflows. 2. Quality may be declining. During the same period that AI adoption has increased, median pull request sizes have nearly doubled. Increases in PR size can serve as an early indicator of technical debt, as higher code volumes generally correlate with increased complexity and potential for bugs. This trend can also introduce additional friction in the review process, as more lines of code generated means more lines of code to review. 3. Some aspects of developer experience are declining. The Developer Experience Index DXI dropped from 67 to 65 over four quarters. AI is improving some aspects of the developer experience—documentation quality, code maintainability, onboarding speed—while creating new friction in others: larger PRs, slower reviews, less incremental delivery. In aggregate, the net effect is currently negative. Velocity metrics alone will tell you things are improving. Developer experience metrics will tell you whether that’s actually true . 4. AI is making codebases easier to understand, but it’s also making the code it generates harder to trust. The Q2 data highlights a striking divergence between two historically correlated software quality metrics. Specifically, from Q1 2026, Code Maintainability improved by 3.8%, whereas Change Confidence decreased by 6.1%. Code Maintainability indicates how easily developers can understand the codebase, while Change Confidence measures their trust that modifications won’t cause production failures. Traditionally, highly maintainable code results in higher confidence when making changes. However, this data reveals a new tension: although AI helps developers understand the code in front of them, they exhibit less trust in the code they are pushing to production. 5. Saved time isn’t converting into innovation. AI users are now saving an estimated 4 to 6 hours per week. However, the innovation ratio, defined as the percentage of time spent on building new features versus maintenance and overhead, has remained flat over the same period of study. This flat trend indicates that the time saved by AI is not currently converting into increased capacity for creating new value. Leaders should keep a close eye on this metric over time. Ideally, innovation ratio will increase as AI frees up engineers to work on more new features. 6. AI spend is accelerating faster than outcomes. Median quarterly organizational AI spend climbed from ~$1.5K to ~$44K over four quarters. Tech-sector spend increased nearly 28x. These numbers will draw scrutiny. Leaders who cannot connect this investment to downstream outcomes feature velocity, innovation ratio, quality may face increasingly difficult budget conversations in the back half of 2026. What this means for leaders The Q2 2026 data indicates that the industry is shifting from base AI deployment to evaluating concrete return on investment. As AI expenditures accelerate, engineering leaders must shift their focus from simply acquiring AI tools to optimizing the surrounding development pipelines and resolving systemic bottlenecks. To achieve true ROI, leaders must ensure that saved hours are reinvested into product innovation rather than absorbed by existing organizational friction. To explore the full data and benchmark your team against 500+ organizations on measures of throughput, quality, and AI tooling cost, download the full report here. https://getdx.com/report/state-of-ai-impact-in-engineering-q2-report/ That’s it for this week. Thanks for reading. -Justin