AI adoption among software development professionals reached 90% in Google Cloud’s 2025 DORA report, yet only 16.2% of respondents in the same research deploy on demand, with multiple deploys per day. That gap between tool adoption and delivery cadence runs through most DevOps statistics in 2026.
Kubernetes now runs in production for 82% of container users, up from 66% in 2023, according to the CNCF Annual Cloud Native Survey. The U.S. Bureau of Labor Statistics projects employment of software developers, quality assurance analysts, and testers to grow 10% from 2025 to 2035. The figures below draw on DORA delivery benchmarks, CNCF and Stack Overflow surveys, JetBrains agent data and BLS pay records, each dated to the survey or release it comes from.
Key Takeaways
- Elite delivery teams recover failed deployments in less than one hour, keep the change fail rate at 5% and make up 19% of respondents in the 2024 performance clusters.
- Only 9.4% of 2025 DORA respondents move a change from commit to production in less than one hour.
- In DORA’s 2024 model, a 25% increase in AI adoption was associated with a 7.2% drop in delivery stability.
- Among Stack Overflow respondents, agent use rose from 31% in the 2025 survey to 59% in an April 2026 pulse survey.
- Claude Code reached around 39% of professional developers at work in May to July 2026, up from 18% in January 2026, JetBrains reports.
- Cultural changes with the development team are the top cloud native challenge, cited by 47% of CNCF respondents.
Editor’s Choice
- Developers using AI tools: 79% of respondents in 2025, up from 44% in 2023, in the Stack Overflow survey.
- Kubernetes in production: 82% of container users in 2025.
- On-demand deployment: 16.2% of 2025 DORA respondents.
- Software developer jobs: 1,905,400 U.S. jobs for software developers, quality assurance analysts, and testers in 2025.
- Median developer pay: $135,980 for U.S. software developers in May 2025.
- GitHub Actions usage: 11.5 billion minutes in public projects in 2025.
DORA Software Delivery Performance Benchmarks
- Four performance clusters emerged from the 2024 survey data: elite teams, with lead times of less than one day, at 19% of respondents, high at 22%, medium at 35% and low at 25%.
- Elite teams have a change lead time of less than one day and recover failed deployments in less than one hour.
- Low performers take between one month and six months to ship a change and record a 40% change fail rate.
- Compared with low performers, elite performers show 127x faster lead time, 182x more deployments per year, an 8x lower change failure rate and 2,293 times faster failed deployment recovery.
- Medium performers post a 10% change fail rate, below the 20% recorded for high performers.
| Performance level | Change lead time | Deployment frequency | Change fail rate | Failed deployment recovery time | Share of respondents |
|---|---|---|---|---|---|
| Elite | Less than one day | On demand (multiple deploys per day) | 5% | Less than one hour | 19% |
| High | Between one day and one week | Between once per day and once per week | 20% | Less than one day | 22% |
| Medium | Between one week and one month | Between once per week and once per month | 10% | Less than one day | 35% |
| Low | Between one month and six months | Between once per month and once every six months | 40% | Between one week and one month | 25% |
Source: DORA, Accelerate State of DevOps 2024
About This Data
These figures are compiled from 13 sources, including the U.S. Bureau of Labor Statistics, two Google Cloud DORA research reports, and developer surveys from Stack Overflow, CNCF and JetBrains. Publication dates run from 2024 to September 2026.
Only official statistics, survey publishers’ own pages, and company platform data qualified. Figures are reviewed on a rolling basis and updated when sources publish new editions.
What are the four DORA metrics?
DORA measures throughput with lead time for changes, deployment frequency, and failed deployment recovery time, and it measures instability with change failure rate. The 2025 framework also counts rework rate as a second instability factor. Together they describe how fast changes reach users and how often they break.
Deployment Frequency Statistics
- 16.2% of 2025 DORA respondents deploy on demand, with multiple deploys per day.
- The largest group, 31.5%, deploys between once per week and once per month.
- 22.7% of respondents deploy at least once per day when the on-demand and hourly-to-daily bands are combined.
- 3.6% deploy fewer than once per six months.
- In the 2024 CNCF survey, 29% of organizations release code multiple times a day, up from 23% in 2023.
View the data
| Deployment frequency | Value |
|---|---|
| Between once per week and once per month | 31.5% |
| Between once per day and once per week | 21.9% |
| Between once per month and once every six months | 20.3% |
| On demand (multiple deploys per day) | 16.2% |
| Between once per hour and once per day | 6.5% |
| Fewer than once per six months | 3.6% |
How often do high-performing DevOps teams deploy?
In DORA’s 2024 clusters, elite teams deploy on demand, with multiple deploys per day, and keep their change fail rate at 5%. In the 2025 survey, 16.2% of respondents report that on-demand cadence. Weekly to monthly release cycles remain the largest single group, though not a majority of teams.
Recent Developments
- September 30, 2026: Stack Overflow reported that agent use rose from 31% in its 2025 survey to 59% in an April 2026 pulse, with Claude Code usage increasing from 41% to 55%.
- September 7, 2026: CNCF and SlashData estimated that China has approximately 1.75 million cloud native developers, within a global community of 19.9 million.
- August 27, 2026: The Bureau of Labor Statistics last modified its Occupational Outlook Handbook page for software developers, quality assurance analysts, and testers on August 27, 2026. The handbook projects 10% growth for software developers, quality assurance analysts, and testers from 2025 to 2035, with about 106,100 openings a year.
- August 2026: JetBrains reported that 90% of professional developers used AI coding agents at work at least weekly in May to July 2026, and 68% used them daily.
- August 2026: JetBrains measured Codex adoption at work rising from 3% in January 2026 to 16% in May to July 2026.
Lead Time for Changes Statistics
- 31.9% of 2025 DORA respondents report a lead time between one day and one week, the largest band.
- 9.4% move changes from commit to production in less than one hour.
- 24.4% ship changes in less than one day when the under-one-hour group is included.
- 56.4% of respondents reach production within one week.
- 2% report a lead time of more than six months.
View the data
| Lead time for changes | Value |
|---|---|
| Between one day and one week | 31.9% |
| Between one week and one month | 28.3% |
| Less than one day | 15% |
| Between one month and six months | 13.2% |
| Less than one hour | 9.4% |
| More than six months | 2% |
Change Failure Rate and Recovery Time Statistics
- 26% of 2025 DORA respondents report a change failure rate between 8% and 16%, the most common band.
- 8.5% keep failures at 0% to 2% of changes, while 5.9% see failures on more than 64% of changes.
- 36.2% of respondents report a change failure rate of 8% or lower.
- 21.3% restore service after a failed deployment in less than one hour.
- 56.5% recover within one day, while 1% take more than six months.
View the data
| Change failure rate | Value |
|---|---|
| 0%-2% | 8.5% |
| 2%-4% | 8.1% |
| 4%-8% | 19.6% |
| 8%-16% | 26% |
| 16%-32% | 19.5% |
| 32%-64% | 12.5% |
| >64% | 5.9% |
Failure rates cluster in the middle bands, so recovery speed decides how much a failed change costs users. The recovery distribution below shows most teams restore service within a day, while a small tail takes weeks.
| Failed deployment recovery time | Share of respondents (%) | Cumulative share (%) |
|---|---|---|
| Less than one hour | 21.3 | 21.3 |
| Less than one day | 35.3 | 56.5 |
| Between one day and one week | 28 | 84.5 |
| Between one week and one month | 9.4 | 93.9 |
| Between one month and six months | 4.9 | 98.8 |
| More than six months | 1 | 100 |
Source: DORA, State of AI-assisted Software Development 2025
CI/CD and Release Automation Statistics
- In the 2024 CNCF survey, 71% of organizations check in code multiple times per day, up from 52% in 2023.
- Among 2024 CNCF survey takers, 38% automate 80% to 100% of their releases.
- The average share of automated releases rose from 56.5% in 2023 to 59.2% in 2024.
- Developers used 11.5 billion GitHub Actions minutes in public projects in 2025, up 35% from 8.5 billion in 2024.
- Developers merged 43.2 million pull requests a month on GitHub in 2025, up 23% year over year.
View the data
| Practice | 2023 (%) | 2024 (%) |
|---|---|---|
| Check in code multiple times a day | 52% | 71% |
| Release code multiple times a day | 23% | 29% |
| Average share of releases automated | 56.5% | 59.2% |
Kubernetes and Cloud Native Adoption Statistics
- In the CNCF survey, 98% of surveyed organizations have adopted cloud native techniques.
- 59% say “much” or “nearly all” of their development and deployment is cloud native, while 10% are in early stages or not using it.
- 66% of organizations hosting generative AI models use Kubernetes for some or all inference workloads.
- Only 7% of organizations deploy AI models daily, and 44% do not yet run AI or machine learning workloads on Kubernetes.
- 58% of cloud native innovators use GitOps principles extensively, compared with 23% of adopters.
- Cultural changes with the development team lead the list of challenges at 47%.
View the data
Why it matters: CNCF found that cultural change with the development team, cited by 47% of respondents, now outranks lack of training and security at 36% each. Container platforms are now standard; the harder work sits in team structure and ownership, which tooling budgets alone do not fix.
Kubernetes adoption tracks the wider shift in cloud computing usage. Inference clusters also add to the server load inside the facilities covered in data center statistics.
DevOps Statistics on Docker, Terraform and Tool Usage
- Among surveyed developers, Docker use rose from 35.0% in 2020 to 71.1% in 2025.
- Kubernetes use grew from 8.5% in 2019 up to 28.5% in 2025.
- Terraform use moved from 6.2% in 2020 to 17.8% in 2025.
- Jira held steady at 46.4% in 2025 against 47.7% in 2020.
- GitLab recorded 35.6% in 2025, while AI tools Cursor (17.9%) and Claude Code (9.7%) made their initial splash in enterprise environments during 2025.
| Tool | Earlier year | Earlier share (%) | 2025 share (%) |
|---|---|---|---|
| Docker | 2020 | 35.0 | 71.1 |
| PostgreSQL | 2018 | 32.9 | 55.6 |
| Jira | 2020 | 47.7 | 46.4 |
| Kubernetes | 2019 | 8.5 | 28.5 |
| Terraform | 2020 | 6.2 | 17.8 |
Source: Stack Overflow Developer Survey 2025
Platform Engineering Statistics
- In the 2025 DORA research, 90% of organizations have adopted at least one platform.
- 88% of backend developers work with at least one form of infrastructure standardization, up from 80% six months earlier.
- The share of developers working without formalized DevOps or platform practices fell from 20% to 12%.
- The global cloud native developer population reached 19.9 million in Q1 2026, roughly 39% of all developers worldwide.
- The Backstage internal developer portal project ranks #5 among CNCF projects by velocity.
| Metric | Earlier period | Earlier reading | Q1 2026 |
|---|---|---|---|
| Backend developers with infrastructure standardization (%) | Six months earlier | 80 | 88 |
| Backend developers classified as cloud native (%) | Q1 2025 | 49 | 52 |
| Developers without formal DevOps or platform practices (%) | Previous reading | 20 | 12 |
| Cloud native developers (millions) | Q3 2025 | 15.6 | 19.9 |
Source: CNCF and SlashData, State of Cloud Native Development Q1 2026
Key finding: In research with SlashData, CNCF counted 19.9 million cloud native developers in Q1 2026, up from 15.6 million in Q3 2025, a 28% increase in six months.
AI Adoption in Software Delivery
- Among Stack Overflow respondents, AI tool use grew from 44% in 2023 to 62% in 2024 and 79% in 2025.
- 84% of 2025 respondents use or plan to use AI tools, and 51% of professional developers use them daily.
- DORA’s 2025 adoption figure of 90% marked a 14% increase from the prior year, with a median of two hours a day spent working with AI.
- 65% of DORA respondents rely heavily on AI for software development.
- Developers trust AI output less than they use it, with 46% distrusting its accuracy against 33% who trust it.
View the data
| Survey year | Value |
|---|---|
| 2023 | 44% |
| 2024 | 62% |
| 2025 | 79% |
Most coding assistants run on large language models, one branch of the wider machine learning market. Adoption describes who touches AI at all. The task data below, from DORA’s 2024 report, shows where that use concentrates in day-to-day delivery work.
View the data
| Task | Value |
|---|---|
| Code writing | 74.9% |
| Summarizing information | 71.2% |
| Code explanation | 62.2% |
| Code optimization | 61.3% |
| Documentation | 60.8% |
| Test writing | 59.6% |
| Debugging | 56.1% |
| Data analysis | 54.6% |
- Code writing leads AI task reliance at 74.9% of respondents whose jobs include the task, followed by summarizing information at 71.2%.
AI Coding Agent Adoption Statistics
- In the 2025 Stack Overflow survey, 52% of respondents either do not use agents or stick to simpler AI tools, and 38% have no plans to adopt them.
- In the United States, Claude Code adoption at work reached 47% in May to July 2026, JetBrains reports.
- GitHub Copilot adoption at work fell from 29% a year earlier to 21% in May to July 2026.
- Cursor adoption slipped from 18% in January to 12% in May to July 2026.
- JetBrains says about 9% of developers worldwide use its own AI tools in IDEs or Junie at work.
View the data
| AI coding tool | Value |
|---|---|
| Claude Code | 39% |
| GitHub Copilot | 21% |
| Codex | 16% |
| Cursor | 12% |
| JetBrains AI | 9% |
| Google Antigravity | 6% |
JetBrains sells AI coding tools of its own, so its adoption figures are a vendor’s measurement; the Stack Overflow pulse points the same way on agent growth.
AI Impact on Delivery Stability and Trust
- In DORA’s 2024 model, a 25% increase in AI adoption was associated with a 7.5% increase in documentation quality and a 3.4% increase in code quality.
- The same increase was associated with a 1.5% drop in delivery throughput.
- 30% of 2025 DORA respondents trust AI “a little” or “not at all,” while 24% report “a great deal” or “a lot” of trust.
- Among Stack Overflow respondents, 66% cite AI solutions that are “almost right, but not quite,” and 45% say debugging AI-generated code takes more time.
| Outcome | Estimated change for a 25% rise in AI adoption (%) |
|---|---|
| Documentation quality | 7.5 |
| Code quality | 3.4 |
| Code review speed | 3.1 |
| Approval speed | 1.3 |
| Delivery throughput | -1.5 |
| Code complexity | -1.8 |
| Delivery stability | -7.2 |
Source: DORA, Accelerate State of DevOps 2024
Worth noting: Among Stack Overflow respondents, only 3% report “highly trusting” AI output, and experienced developers show the highest “highly distrust” rate at 20%. Review and testing capacity, not code generation, sets the limit on how much AI output a team can safely ship.
Software Developer Employment and Job Outlook
- U.S. software developers held about 1.7 million jobs in 2025, and software quality assurance analysts and testers held about 187,600.
- The Bureau of Labor Statistics projects software developer employment to rise from 1,717,800 in 2025 to 1,892,600 in 2035, an increase of 174,700.
- Combined employment for developers, analysts and testers is projected to grow by 185,400 jobs from 2025 to 2035.
- Computer systems design and related services employ 29% of software developers, ahead of finance and insurance and software publishers.
- More than 180 million developers work on GitHub, and over 36 million joined in the past year.
The BLS job counts cover U.S. roles, while the GitHub figure counts accounts worldwide, so the two series measure different populations.
By the numbers: The Bureau of Labor Statistics projects about 106,100 openings a year for software developers, quality assurance analysts, and testers over the decade, with employment growth of 10% from 2025 to 2035. Many of those openings replace workers who retire or change fields, so hiring demand runs ahead of net job growth.
Software Developer Salary Statistics
- U.S. software developers earned a median of $135,980 a year in May 2025, against $50,980 for all occupations.
- The lowest 10% of software developers earned less than $82,460, and the highest 10% earned more than $214,670.
- Software quality assurance analysts and testers earned a median of $104,300.
- Among Stack Overflow respondents, median pay for cloud infrastructure roles rose 6.7% between 2024 and 2025, compared with 8.4% for system administrators.
- Software publishers pay developers the highest median of these industries at $164,550.
View the data
| Industry | Value |
|---|---|
| Software publishers | $164,550 |
| Manufacturing | $136,330 |
| Management of companies and enterprises | $135,680 |
| Finance and insurance | $135,460 |
| Computer systems design and related services | $132,050 |
How much do DevOps and software engineers earn?
U.S. software developers earned a median of $135,980 in May 2025, with software publishers paying $164,550. BLS does not break out DevOps engineers as a separate occupation, so the software developer series is the closest official benchmark for DevOps pay.
Developer Job Satisfaction and Work Setting
- In the 2025 Developer Survey, 24.5% of respondents feel happy at work, 47.1% feel complacent, and 28.4% feel unhappy.
- 32.4% work fully remotely, 17.9% work on-site, and 37.1% work in one of two hybrid models.
- 12.6% choose when to go into the office.
- 82% have some degree of remote flexibility in 2025, slightly above the 80% combined total recorded in 2024.
View the data
| Job satisfaction | Value |
|---|---|
| Complacent | 47.1% |
| Not happy | 28.4% |
| Happy | 24.5% |
Conclusion
AI use reached 90% of software development professionals in DORA’s 2025 research, while 16.2% of respondents deploy on demand. The gap shows that adoption and delivery speed move on different clocks: tools spread in a year, while release practice, testing, and team culture change slowly. Teams with automated releases and stable platforms are best placed to turn AI output into shipped, working software.
The Bureau of Labor Statistics expects employment of software developers, quality assurance analysts, and testers to grow 10% from 2025 to 2035. The next Stack Overflow and DORA releases will show whether agent use keeps climbing and whether the throughput and stability losses DORA linked to AI in 2024 persist.
