Azul’s 2026 State of Java Survey & Report reinforces the point that Java is still a mainstay of enterprise IT environments and remains trusted for mission critical workloads including the most valuable AI initiatives. 64% said that more than half of their applications or workloads are built with Java or run on a Java Virtual Machine (JVM).
And yet, we are in a period of significant change in which AI vendors are suggesting they could dramatically alter how code is developed and maintained. Despite this potential disruption, the clear message from survey respondents is that they are totally focused on maximising the value of their Java investments and that business-critical systems, including those with the most progressive AI capabilities, are stood up in Java.
One such change, it moving to a vendor neutral platform. Strikingly, 63% globally intend to migrate off their entire Oracle Java estate to a non-Oracle OpenJDK distribution showing a determination not to be dependent on the owner of the Java source code particularly as it pertains to the systems and initiatives they care about most. In this year’s report, there are three other interesting themes that come through in the data: optimising performance, minimising distractions and keeping pace with innovation.
Optimising performance
Survey respondents cite several developments that fall broadly under the banner of optimising Java environments for performance. For workloads where performance matters, including AI and big data initiatives, it’s clear that companies are asking for runtimes that can boost results. One such way, is migrating to the latest versions of the JDK as it enables users to access the benefits of more performance applications and cloud efficiencies. In September 2025,18% were already on JDK 25 just weeks after its release and 37% are on JDK 21.
Finding ways to optimise their cloud environments is also a focus as 74% say that more than 20% of their cloud compute costs come from JVM-based workloads. A similar percentage say they have more than 20% unused compute capacity in their public cloud environments due to overprovisioning to compensate for inconsistent or unpredictable runtime behaviour, as well as slow start-up times and pro-longed warm-up cycles. Taking into account investments in AI, both of these concerns are only escalating. This is a clear drag on efficiency, and respondents are determined to take action by adopting a high-performance Java platform (41%), establishing and tracking metrics for usage and costs (40%), and leveraging cloud provider tools and best practice (38%).
Minimising distractions
With growing pressure to produce functionality and applications that help organisations respond to customers and market opportunities, developers want to use their time productively rather than be distracted by maintenance and operational issues. Unfortunately, the State of Java report suggests 30% of respondents are concerned their teams waste more than half their time chasing false positives and 36% are dealing with common vulnerabilities and exposures (CVEs) in JVM workloads every week. Respondents also admit they are struggling with the hidden costs of technical debt, and 63% say dead and unused code affects their team’s productivity. With the vast number of lines of Java code built up in organisations around the world, it should be no surprise that some of the code is not being used. Developers must carefully consider the implications of integrating such code into applications or leaving it to lie dormant in their IT infrastructures, because it could expose them to unwanted security risks. Equally, they might not need to expend effort on unused code if it does not contain security vulnerabilities. Filtering what code needs to be updated or removed is a painstaking process and can significantly impact developer productivity.
The introduction of new AI tools like Anthropic Claude Mythos in recent news highlight the urgency to reduce this exposure, and eliminate dead and unused code as low-hanging fruit for today’s DevOps teams.
Innovation
Like everywhere else in the technology industry, the Java community is also heavily involved in experimenting with artificial intelligence. Given the preconception that Python is the language of AI, it is striking that so many respondents to the Azul study are using Java for AI development. 31% say more than half of the Java applications they build now include AI functionality and 62% of organisations now use Java to code AI functionality, a figure that is up from 50% last year.
This underscores the importance of Java in the enterprise, and that Java remains vibrant and relevant in this age of AI. It is another reason why developers are keen to be on the latest version of the JDK as it provides access to containerised apps and AI-driven workloads.
Respondents say they use a variety of tools for automatic updates to Java code for patches, new Java versions and vulnerability remediation. 58% turn to OpenAI’s ChatGPT, 51% are using Gemini AI and 32% are using Amazon Q. ChatGPT comes out on top when it comes to using AI code generation tools to create new Java applications. OpenAI leads (52%), followed by Google Gemini Code Assist (43%) and Microsoft Visual Studio IntelliCode (37%), but given how quickly the technology is evolving it won’t be surprising to see this order changing in the future.
Last year, Java celebrated its 30th birthday, but as this year’s State of Java Survey and Report suggests, it shows no sign of slowing down and is reacting quickly to the ever-changing demands of its users. It will be exciting to see how AI shapes future use but make no mistake: Java remains at the heart of many mission-critical processes precisely because it continues to satisfy the need for improved efficiency, productivity and continuous innovation.
