Azul CEO sees Java’s AI future as bright

Azul Systems specializes in providing Java technologies ranging from supported OpenJDK builds to Azul Intelligence Cloud, offering actionable intelligence from Java runtime data. The company maintains both technical and competitive relationships with Oracle, the Java steward. InfoWorld Editor at Large Paul Krill recently interviewed Azul CEO Scott Sellers. The following interview touches on Java’s place in the burgeoning AI space and where Java fits in the world of software today.

InfoWorld: Java is usually not the first language that comes to mind with AI; it usually would be Python. Do you think Java has what it takes to compete in the AI space?

Sellers: When you look at Python’s use with AI and look at what it does,  it’s like glue code, really. There’s not really a lot that Python itself is doing. It’s actually mostly calling to native libraries, or the GPU libraries. The interface is the key thing. There’s no question that Java will, I think, become as popular as Python over time in terms of AI. The more that AI is incorporated into traditional business logic and those things that need to happen at a true application level, the more that enters the sweet spot of Java and the popularity of Java. Python’s very limited in terms of performance and scale and those types of things.

InfoWorld: Where do you see Java headed?

Sellers: I think the change that the community made beginning with Java 9 in this movement to the six-month release cadence and combining the rapid evolution of the Java runtime and the Java platform in general with identified long-term support versions, I think that has gone incredibly well. You remember the days when it took three, four-plus years for a given major version of Java to get released and it was painful. The community has been tremendous in terms of really embracing the philosophy behind the release train methodology and mentality.

As a result of that, what you’re seeing is a really nice pace of innovation coming from the Java platform. We now have the ability to evolve to meet the ever-changing needs of developers. Things like the Foreign Function & Memory API  that came about with Java 22, it’s really important to allow the Java platform to continue to address some of the limitations. You’re seeing more vector APIs and things like that coming. All these things, I think, are contributing to keep the flywheel spinning.

InfoWorld: As far as  the upcoming JDK 23 release, are you familiar with sun.misc.Unsafe methods that are being removed?

Sellers: Yep. It’s a big deal. It’s long overdue. As the name implies, it is very unsafe and it takes away many of the advantages of the Java platform itself, which is fully protected, robust, secure, and difficult to penetrate. Unsafe creates an unsafe spigot that, until APIs and interfaces like the new Foreign Function & Memory API, developers had no choice when they needed to do something outside of Java but to go through this very unsafe interface. It’s been long overdue in terms of cleaning that up and allowing ways for Java applications to interface with non-Java things (such as) GPUs. It definitely closes a security hole. It closes a robustness hole. But It’s going to be, I think, another challenging transition as apps have to do some real work to remove it.

InfoWorld: Can you tell me about Azul Intelligence Cloud and what that means for the company?

Sellers:  Intelligence Cloud Is Azul’s first SaaS offering. The idea behind it is to be able to take information that is inside the running JVMs across an enterprises’s fleet and, in a transparent manner, being able to send data to the Intelligent Cloud SaaS offering, and being able to retain and do interesting analysis and provide what we call actionable intelligence for the users of Intelligence Cloud. There are two main use cases today. The first is vulnerability detection in production. The other has to do with code maintenance and overall modernization initiatives.