KubeCon Europe 2026 in Amsterdam felt less like a conference focused on Kubernetes and more like a glimpse into where the cloud native ecosystem is heading next. The conversations have shifted, and Kubernetes itself is no longer the main attraction. Instead, the focus has moved to the workloads running on top of it, how platforms are evolving, and how teams are adapting to new pressures from AI, virtualization, and ever growing operational complexity. Across the action packed week, a few themes consistently stood out. Cilium is becoming foundational, to the point where it is rarely questioned. Virtual machines are making their way back into the conversation, this time as part of a Kubernetes first strategy. And while AI is everywhere, the community is still figuring out how to operationalize it, with early patterns like agentic systems starting to emerge. It would be impossible to recap everything that happened, but here is our perspective on what it all means. Rejekts Amsterdam Rejekts is one of our favorite “b sides” conferences. This year it was successfully run entirely by volunteers! Duffie Cooley, our Americas Field CTO, was a track host for the main room. There were quite a few presentations at Rejekts that included Cilium or Tetragon including “Unleashing the Tides of Kubernetes Networking by Removing kube-proxy” There were over 220 people from all over the world that arrived early to celebrate this community with us! CiliumCon Three years later, CiliumCon returned to where it started for its seventh edition, at KubeCon + CloudNativeCon Europe in Amsterdam. Mahé Tardy, software engineer at Isovalent co-chaired the event, while Isovalent community team member and Cilium maintainer Bill Mulligan delivered a keynote about why Cisco is betting big on Cilium, eBPF, and Tetragon redefining how networking, security, and observability are built in the multi-cloud world. The keynote also featured a special appearance from a Cisco Switch, and attendees loved seeing a piece of the data center in the cloud native world. Before closing, Martynas Pumputis, software engineer at Isovalent, gave a lightning talk on the latest major advancements in the Cilium ecosystem, highlighting the most promising initiatives which were particularly well received by the attendees. End user talks from organizations like Microsoft, Roche, Etraveli Group, Ledger, and SUSE reinforced that Cilium is a real-world, battle-tested project while talks from contributors and maintainers helped community members stay current with fast-moving developments. Attendees walked away having heard from organizations with production-tested approaches to running Cilium in complex, real-world environments. eBee Quest The evening before KubeCon was reserved for a special event: a small group of Isovalent fans met for a special hands-on lab experience, the eBee Quest. Located some 30 floors above the city, participants sat down with good food and went on a quest to do our labs. Guided by our experts, every participant chose their own path through our map and did lab after lab - while having great food and great conversations alike. It was an amazing opportunity to connect with each other in a relaxed environment with a spectacular view over the city. Highlights of the evening were discussions about how to use Cilium to manage satellite data streams, but also people doing the labs on their mobile devices and even on virtual reality headsets. But the Quest didn’t end that evening! The quest kept going throughout the week and you can join them as part of a virtual quest until the 17th of April! Read more at our blog post The eBee Quest Lands at KubeCon Amsterdam. Has Cilium Become Boring? At KubeCon Europe 2026, Cilium did not feel new, and that is exactly the point. Cilium has quietly crossed an important threshold. It is no longer a “modern alternative” CNI but increasingly the default choice for Kubernetes networking. In many conversations, the question was not whether to adopt Cilium, but what more to do with it once it is already running. That shift is important because when a technology becomes table stakes, differentiation moves up the stack. The value is no longer in replacing kube-proxy or improving raw network performance, but in everything that builds on top of it. That change showed up clearly in the kinds of questions being asked. Teams are revisiting service mesh decisions, often running Cilium alongside Istio but questioning whether a separate control plane is still necessary. Cilium’s built-in service mesh capabilities are no longer treated as experimental. They are increasingly seen as a path toward consolidation. A similar pattern is emerging at the edge. With the deprecation of NGINX Ingress Controller, platform teams are rethinking ingress strategy altogether. If Cilium already handles networking, policy, and observability, extending it to Gateway API starts to feel less like a trade off and more like a natural progression. There is also growing interest in how Cilium applies beyond containers. As Kubernetes platforms expand to include virtual machines, the challenge becomes providing consistent networking across mixed workloads. Cilium is increasingly part of that conversation, positioning itself not just as a CNI but as a unifying layer across environments. None of this is flashy, and that is what makes it notable. There were no major announcements driving this shift, just a steady pattern of adoption and deeper usage. Cilium is becoming boring, and in infrastructure, that usually means it has won its place. Boring also does not mean forgotten, as evidenced by the standing room only crowd at CiliumCon, and the packed two-story auditorium during Cilium’s maintainer track session. The project also got a keynote shoutout from Netlify, who is using Cilium in its cloud native stack, and End User Award winner SNCF, the French railway operator who is using Cilium for high-performance networking and observability. There was also a presentation from Asana about how they built security functionality into their platform with Cilium. Virtual Machines: An Unlikely Hot Topic At KubeCon Europe 2026, virtual machines are quietly reclaiming a place in modern platform architecture. Running VMs on Kubernetes is no longer an experiment for organizations. Platform teams are hearing the same message repeatedly: the business wants to move virtual machine based workloads onto Kubernetes. This is no longer just a theory and it is fast becoming a real operational choice. Networking is central to making this work, and Cilium has become an essential part of the conversation. Platform engineers are no longer asking only about container networking. They are asking how to extend the same capabilities they are used to using for Containers like visibility, policy, and security to virtual machines. Concepts such as the “Kubernetes VPC” and EVPN integration are gaining traction because they provide a consistent networking layer across mixed workloads of Containers and VMs. These discussions are no longer just hypothetical as they are practical and strategic. Platforms based on Open Source tooling remain part of the dialogue as enterprises are thinking beyond the vendor stack. They want flexibility and a networking layer that works seamlessly whether the workload is a container or a VM. Cilium’s expanding capabilities in this space are positioning it as a good foundation for hybrid environments. The broader takeaway is that VMs on Kubernetes are no longer a curiosity. Platform teams are actively exploring how to unify networking, policy, and observability across all workloads. Cilium is emerging not just as a CNI, but as the glue that makes this vision achievable. The conversations at KubeCon reflect a subtle but important shift: virtualization and Kubernetes are converging, and networking is at the center of that evolution. AI Workloads Are Here, But the Community Is Cautious The cloud-native community remains thoughtfully cautious when it comes to AI at KubeCon EU 2026. In past events, AI on Kubernetes was mostly a question mark. People were curious, but no one really knew how it fit into the cloud-native landscape. This year the conversation has changed just a bit. It is no longer about whether AI or ML workloads belong on Kubernetes. It is about how to run them effectively, efficiently, and safely at scale. Agentic AI is emerging as the approach most teams are exploring for running AI workloads in Kubernetes.These agentic systems/solutions drew attention not as research curiosities but as practical tools for orchestrating AI tasks and making AI workloads more manageable at scale on Kubernetes. Separately, maintainers are grappling with the impact of AI-generated contributions on open-source projects. Many are discussing how to maintain quality and consistency while dealing with the influx of “AI slop,” and ironically, some are considering using AI itself to help manage these contributions. NVIDIA reinforced this trend with a strategic announcement. Donating the Dynamic Resource Allocation (DRA) Driver for GPUs to CNCF moves a critical piece of AI infrastructure into open-source, community governance. This allows developers to orchestrate high-performance AI workloads with transparency and flexibility while signaling that open source remains the cornerstone for AI in cloud-native environments. The takeaway is clear. The community has moved from uncertainty to early adoption patterns. Enterprise teams are focused on operationalizing AI workloads, maintainers are exploring how to manage AI-driven contributions at scale, and agentic systems are becoming a tangible framework for doing it. AI is no longer just a future possibility in Kubernetes. Organizations are seeking to best operationalize it, and open-source tools are the foundation that makes it work. Open Source Health Starts With Inclusion Beyond the technology and announcements, KubeCon Europe 2026 once again highlighted that the community is what makes the cloud native ecosystem unique. Open source does not just depend on code, it depends on people, and ensuring new voices can participate and grow is critical to its future. Programs like the Dan Kohn Scholarship and initiatives like Merge Forward are helping make that happen. The Dan Kohn Scholarship: Bringing New Voices to KubeCon Diversity matters more than ever in a world where open source contributors and maintainers are increasingly hard to find. We witnessed this firsthand when nginx was archived live by James Strong during Christine Kim and Nick Young's talk. Cilium, Isovalent's first product, was born as an open source project, and at Isovalent, diversity has always been central to building great products, not just in terms of the people involved, but also the users and companies they serve. The broader our perspective, the better the outcome. That's why Isovalent has sponsored the Dan Kohn Scholarship every year since KubeCon EU 2022 in Valencia, giving people from underrepresented backgrounds the opportunity to attend one of tech's biggest events. A chance to connect with companies, fellow contributors, ambassadors, and peers, and to learn through talks and hands-on workshops. Ahlem Jaouadi, DevOps Engineer and social media manager of the Merge Forward initiative, was one of this year's Dan Kohn Scholarship recipients. She shared her excitement on a LinkedIn Post before her first KubeCon and we asked her what she took away from the conference: The event was massive. There was so much happening that I know I barely scratched the surface. If only I could have typed k scale ahlem --replicas=3 and sent a few copies of myself out into the crowd, I would have. I met so many people, all incredibly welcoming and genuinely willing to help. The perfect occasion to connect with Merge Forward members too. I also stumbled upon open source projects I am now really excited to dig into. And I may have spent a suspicious amount of time hunting down signed books, because yes, I am absolutely a book lover. Everything about it was amazing. I cannot wait to come back next year! Merge Forward: Building Space for Everyone in Cloud Native In case you missed it, Merge Forward is the CNCF initiative bringing together 7 community groups: Deaf and Hard of Hearing, Blind and Visually Impaired, Stuttering and Speech Diversity, Women in Cloud Native, Deep Roots (Black, Indigenous, and People of Color), Friends of Dorothee (LGBTQ+), and Neurodiversity. If you're looking to connect with like-minded people, this is your chance. Between the Project Pavilion kiosk hosting all seven groups on rotation every hour and the community hub sponsored by Google with more than 10 sessions, including a Sign Language Course, Women's gathering, Deep Roots gathering, and Neurodiversity gathering, there was something for everyone to learn from, contribute to, and connect around. It's also a real opportunity to nurture the next generation of open source contributors, and maybe tomorrow's maintainers. Below is an interview with Anastasiia Gupska, the first deaf CNCF Ambassador and a true role model in the open source community, alongside Donia Chaiehloudj, software engineer at Isovalent and representative of the Merge Forward initiative. Summary KubeCon Europe 2026 made one thing clear: the cloud native ecosystem continues to evolve quickly, and the conversations are getting more practical with every event. From foundational technologies like Cilium to emerging patterns around AI and virtualization, it is an exciting time to be building in this space. If you want to go deeper on Cilium, join us for our upcoming Cilium Up & Running webinar on April 9 And with that, we are already looking ahead. See you at KubeCon North America in November!