Kenny Johnston steps into the role of Chief Product Officer at Azul during a transformative period for enterprise software, bringing a sophisticated blend of experience in agentic AI and DevOps infrastructure. Having previously led product and technology at Luciq, where he pioneered agentic AI observability, and served as a key leader at GitLab, Johnston is uniquely positioned to bridge the gap between traditional Java environments and the high-speed demands of modern AI. This conversation explores how Java is evolving to meet the intense pressure of AI workloads, the strategic importance of runtime optimization over hardware expansion, and Azul’s roadmap for securing and modernizing the backbone of the world’s most critical enterprise systems.
The interview explores the intersection of high-performance Java and generative AI, the technical hurdles of throughput and latency in the JVM, and the strategic direction of Azul’s expanded portfolio following its recent acquisitions.
Your background spans agentic AI observability at Luciq and DevOps infrastructure at GitLab. How do these distinct experiences influence your approach to leading Azul’s AI-first Java strategy?
My journey has always been about helping engineering teams navigate massive technological shifts with confidence. At Luciq, I saw firsthand the friction that occurs when agentic AI is deployed into production, specifically through the development of capabilities like SmartResolve and Agentic Mode, which required a deep understanding of how AI agents interact with underlying services. Moving back further, my time at GitLab allowed me to oversee the backbone of enterprise development, from CI/CD to Infrastructure as Code and incident response, which taught me that developers need seamless, automated tools to move fast without breaking things. Even earlier, during my tenure at Rackspace, we managed to more than triple annual revenue by focusing on high-performance private clouds, earning us the title of HPE’s Global Service Provider Partner of the Year. I’m bringing all of these lessons to Azul because Java isn’t just a legacy language; it is the engine powering 37 percent of the Fortune 100 and half of the world’s most valuable brands. My focus is on infusing that “DevOps-centric” mindset into the Java runtime, ensuring that the infrastructure is as agile and observant as the AI agents it now supports.
Java has long been the standard for enterprise systems, but AI adoption is creating new types of pressure. What are the specific performance bottlenecks that modern AI workloads are exposing within the JVM?
The reality is that agentic workflows are calling Java services at volumes we’ve never seen before, often at a frequency far higher than traditional human-triggered workloads. When these services sit directly in the critical path of an AI interaction, any delay becomes a glaring failure in user experience. We are seeing major throughput bottlenecks and the dreaded garbage collection pauses that can cause a “stutter” in response times, making an AI feel sluggish rather than intelligent. Furthermore, JVM warm-up delays and heavy memory overhead mean that the first few interactions with an AI service can be frustratingly slow. In the world of financial trading—where we power the top 10 companies globally—a millisecond can be the difference between profit and loss, and AI is now bringing that same level of urgency to every other enterprise sector. We are moving away from a world where you can simply wait for a system to “spin up”; the demand for immediate, consistent performance is now the baseline requirement.
When enterprises face these performance hurdles, the instinctive reaction is often to increase cloud capacity. Why is Azul advocating for a different approach with the Azul Prime runtime?
Throwing more compute at a performance problem is a reflex that has become prohibitively expensive in the era of AI. It’s a “brute force” solution that masks inefficiency rather than solving it, leading to spiraling cloud costs that can stifle innovation. With Azul Prime, we have engineered the runtime to address these bottlenecks directly at the source, allowing the JVM to handle more work with the same amount of underlying hardware. By optimizing how the runtime manages memory and executes code, we provide a more predictable environment that eliminates those erratic spikes in latency without requiring a massive increase in server footprint. It’s about being smarter with the resources you already have, ensuring that as AI workloads expand, your infrastructure costs don’t grow at the same exponential rate. This efficiency is vital for the 50 percent of Forbes Top 10 World’s Most Valuable Brands that rely on us to keep their operations lean and their applications lightning-fast.
Following the acquisition of Payara in late 2025, how does the integration of Payara Micro and Payara Server fit into your broader roadmap for application modernization?
The addition of Payara to our portfolio was a strategic move to ensure we can support the entire lifecycle of a Java application, especially as enterprises look to modernize their stacks for AI compatibility. Payara Micro and Payara Server provide the flexibility needed for cloud-native deployments, which is exactly where most AI-driven modernization is happening today. When we pair these tools with Azul Intelligence Cloud, we create a powerful ecosystem that allows developers to identify exactly where their code needs updates to handle modern throughput demands. Our goal is to provide a seamless bridge from older, monolithic systems to agile, AI-ready microservices without sacrificing the stability Java is known for. This isn’t just about moving to the cloud; it’s about transforming those services into highly scalable, automated components that can participate in complex agentic workflows. We want to ensure that whether you are a financial giant or a growing tech firm, your path to modernization is clear, secure, and performance-optimized.
Security and patch velocity are often cited as major concerns as AI expands the enterprise attack surface. How is the Azul Core platform evolving to protect these increasingly complex environments?
As AI expands the attack surface, the window of time that organizations have to patch vulnerabilities is shrinking, making “patch velocity” a critical metric for any security team. Azul Core is specifically designed to narrow these security gaps by simplifying the deployment of updates across the entire enterprise estate, ensuring that systems aren’t left vulnerable to exploits. We recognize that in an AI-first world, an unpatched JVM is a massive liability, as automated agents can inadvertently expose or interact with insecure services at scale. Our focus is on providing the automation and reliability required to keep these systems current without disrupting the developer’s workflow or the application’s uptime. By hardening the runtime and streamlining the delivery of security fixes, we allow enterprises to focus on building new AI capabilities rather than constantly looking over their shoulders at potential infrastructure threats. It’s about building a foundation of trust so that the next phase of growth is built on solid, secure ground.
What is your forecast for the role of Java in the enterprise as agentic AI becomes the standard for digital interactions?
I believe we are entering a “Java Renaissance” where the language will prove itself once again as the only platform capable of providing the scalability and security that global enterprises demand for AI infrastructure. Over the next few years, we will see a shift where the JVM is no longer viewed as a passive container for code, but as an active, intelligent partner that optimizes itself in real-time based on the specific demands of AI workloads. We will move beyond simple performance tuning and toward a reality where the runtime, enhanced by insights from platforms like Azul Intelligence Cloud, automatically adjusts to mitigate security risks and throughput bottlenecks before they even impact the user. Java has survived and thrived through every major tech cycle because of its resilience, and with an AI-first vision, it will remain the cornerstone of the world’s most valuable brands. The companies that succeed will be those that stop treating their infrastructure as a commodity and start treating it as a strategic advantage in the race for AI dominance.
