A new Java productivity report shows rapid growth in developer use of AI tools, now at over 75 percent, though 12 percent of respondents said their companies disallow their use.

The report is from JRebel by Perforce, the JRebel product being a Java developer productivity tool.

In 2025, 731 developers, team leads and executives working with Java responded to the survey, up from 440 in the equivalent 2024 report. How respondents were selected is not stated, though we might expect some bias towards Perforce customers and users of the JRebel tools.

The 2024 report stated that “only a small subset of respondents (8 percent) reported using AI tools to assist in writing code,” whereas the new report tells us that “Only 12 percent of respondents said they don’t use AI tools, and an additional 12 percent said their companies don’t allow the use of AI tools;” a confusing statement but with the implication that over 75 percent of these teams now use AI. A chart then tells us that 17 percent “don’t use AI tools.”

Unfortunately this report has a number of these inconsistencies suggesting that Perforce perhaps rushed its publication; however we do not doubt the rapid growth in AI tooling adoption.

GitHub CoPilot is the top choice of respondents followed by ChatGPT. The top usage is for code completion (60 percent), followed by refactoring (39 percent) and error detection (30 percent).

The survey asked respondents to name their primary JDK distribution, though the percentages sum to more than 100 percent so this question may have been treated more as one asking about JDK distributions in use – more evidence of a sloppy survey. Still, we learn that only 42 percent named Oracle Java, and that at 23 percent Amazon Corretto is nearly at the same level as Adoptium, from the Eclipse Foundation working group. Despite the sometimes uneasy relationship between AWS and open source, the cloud giant is a key player in the open source community. Organizations deploying to AWS may feel that Corretto will be the best supported version of Java, and have the reassurance that Amazon states that it uses Corretto for its own production services.

AWS is indeed the favoured cloud provider for these respondents, with 51 percent usage, following by Microsoft Azure at 27 percent and Google at 19 percent. Some 20 percent though reported not using a cloud provider at all.

What are the pain points? One is increased startup time for microservices. 44 percent reported some increase, presumably over the last year or so, with 10 percent experiencing an increase of 25 percent or more. Startup time impacts application performance after redeployment, and in cases where applications scale by starting more service instances.

We asked Perforce about the data discrepancies and it said its team had updated the report to correct the errors, and thanked DevClass for pointing them out.