Guest: Simon Ritter (LinkedIn)
Company: Azul 
Show Name: 2026 Predictions
Topic: Cloud Native

As enterprise AI demands accelerate in 2026, the programming language wars are being decided not by popularity but by scalability. While Python dominates data science and AI development, its architectural limitations make it unsuitable for the internet-scale workloads that enterprises require—and that’s where Java‘s three decades of performance optimization are positioning it as the essential glue holding enterprise AI systems together.

Simon Ritter, Deputy CTO at Azul, believes 2026 will be the year Java solidifies its role as the foundation for enterprise AI applications. His predictions center on a fundamental truth: building AI prototypes and scaling AI systems to serve millions of users are entirely different challenges.

Python’s Scalability Problem

The issue with Python in enterprise environments comes down to architecture. “Python has a thing called the GIL, the Global Interpreter Lock,” Ritter explains. “What that essentially means is that you can only do one thing at a time. If you’re trying to handle lots of connections—lots of people trying to use your system at the same time—then you’re going to run into a problem.”

Java, by contrast, has spent 30 years solving concurrency and scalability challenges. The platform’s multi-threading capabilities and recent innovations like virtual threads—which map multiple Java threads to a single platform thread—enable enterprises to handle massive workloads efficiently. This isn’t theoretical: Twitter’s migration from Ruby on Rails toward Scala (running on the JVM) for its backend services is an early example of how moving to a JVM-based stack can deliver the performance and scalability needed by high-growth companies.

AI Will Accelerate Java Workloads

Ritter predicts that AI will actually drive increased demand for Java compute resources, even though most AI development happens in Python. The reason is practical: enterprises won’t build AI systems from scratch. Instead, they’ll layer AI capabilities onto existing Java applications where large datasets and user interactions already exist.

“If you’re going to build more applications on top of that, that’s going to drive a much greater need to run those Java workloads, because we’re going to have more extraction of information, more extraction of value from it, and more data that we need to process,” Ritter says.

The Vibe Coding Reality Check

AI-assisted coding tools like GitHub Copilot and Claude are generating significant hype, but Ritter cautions against expecting them to replace enterprise developers. The fundamental problem is that natural language is inherently ambiguous—unsuitable for mission-critical applications that must perform exactly as specified.

Ritter offers a simple example: “The chicken is ready to eat” could mean a hungry chicken or a cooked dinner, both valid interpretations with very different meanings. This ambiguity makes pure vibe coding dangerous for enterprise applications.

However, Ritter does see AI making developers more efficient through predictive coding in IDEs and automated component generation. “AI-driven coding is going to make developers more efficient. It’s not going to do away with them completely,” he notes.

From Cloud Migration to Java Modernization

Many organizations that embraced cloud migration are discovering their costs haven’t decreased as expected. The culprit is often a “lift and shift” approach that moves monolithic applications to the cloud without architectural changes.

“You end up with a situation where you’re over-provisioning, keeping instances running far too long, and it actually costs you more than if you ran it in your data center,” Ritter explains.

The solution is Java modernization: breaking monolithic applications into microservices that can scale dynamically based on demand. This approach allows organizations to spin up additional instances of specific bottleneck components rather than scaling entire applications.

Runtime performance optimization also plays a crucial role. For example, improving the efficiency of the Java Virtual Machine running a Kafka cluster can reduce the number of nodes needed to handle the same transaction volume, directly cutting cloud costs.

Converging DevOps and FinOps

Ritter’s final prediction addresses organizational structure: DevOps teams focused on deployment and performance will increasingly merge with FinOps teams focused on cost optimization. This convergence will center on return on investment per application as a core metric.

“When you’re deploying new applications or running existing ones, how can we get the most from that in terms of performance—reducing cloud costs and making sure you’re getting the best return on your investment?” Ritter asks.

For Azul, 2026 brings expanded focus beyond Java runtimes. The company’s recent acquisition of Payara positions it to serve the large Jakarta EE (Enterprise Java) market with optimized runtime solutions that reduce both licensing costs and infrastructure requirements.

As economic pressures continue, the opportunity to demonstrate measurable cost savings while maintaining service levels will separate effective IT leaders from the rest. Java’s proven scalability, combined with modern architectural patterns and performance optimization, offers a clear path to that goal.