A new global survey of more than 2,000 Java professionals suggests enterprises are increasingly using Java to support AI workloads, while accelerating migration away from Oracle’s commercial Java distribution in response to pricing and licensing concerns.
The 2026 State of Java Survey & Report, released by Azul, found that 62 per cent of organisations now use Java to build AI-related functionality, up from 50 per cent the previous year. The data indicates that Java is being integrated into production AI environments rather than confined to legacy enterprise systems.
According to the survey, 31 per cent of respondents said more than half of the Java applications they develop now include AI features. Respondents cited long-term support for modern Java versions, built-in security features, observability and support for large data access as key requirements for Java to remain competitive in AI-enabled development environments. Integration with large language models was also identified as an emerging priority.
At the same time, concern over Oracle’s employee-based Java pricing model appears to be driving migration decisions. The survey found that 92 per cent of respondents are concerned about Oracle Java pricing, while 81 per cent have migrated, are migrating, or plan to migrate at least part of their Oracle Java deployments to a non-Oracle OpenJDK distribution. Sixty-three per cent indicated plans to migrate their entire Java estate.
Cost was cited as the primary driver for migration (37 per cent), followed by a preference for open-source distributions (31 per cent), uncertainty around licensing changes (29 per cent) and perceived audit risk (26 per cent). Twenty-one per cent of respondents reported having been subject to an Oracle Java audit.
Cloud cost optimisation also featured prominently in the findings. Ninety-seven per cent of respondents said they had taken steps to reduce public cloud expenditure, with 41 per cent identifying the use of high-performance Java runtimes as one of their cost-reduction strategies. Among organisations where at least 90 per cent of applications run on Java, the use of high-performance Java platforms rose to 81 per cent.
Despite these optimisation efforts, 74 per cent of organisations reported more than 20 per cent unused compute capacity in their public cloud environments, suggesting continued overprovisioning linked to performance variability and runtime inefficiencies.
The survey also highlighted persistent productivity challenges. Sixty-three per cent of respondents said dead or unused code affects team productivity, while 56 per cent reported dealing with Java-related CVEs on a daily or weekly basis. Thirty per cent said their teams spend more than half their time investigating false-positive vulnerability alerts, often related to code paths not used in production.
The findings reflect a broader shift in enterprise Java usage: from legacy application support toward AI integration and cloud optimisation, alongside structural reassessment of licensing models and operational efficiency.
You can read the full report here.
