5 Things KubeCon Atlanta Revealed About Cloud-Native in 2025

Quite a few people mentioned that KubeCon didn’t feel larger this year (flight issues were extreme), but it felt more connected. And yes, the WiFi being down was a whole situation, and yes, there’s definitely a joke in there about how many extremely technical Kubernetes engineers it takes to get conference WiFi working. Apparently, even that wasn’t enough.

But the outage did something great. Without the ability to show off fancy demos, we got the chance to kick off real conversations about the challenges and solutions people face day to day.

Those conversations quickly turned to where we see things heading next. Here’s what we heard.

1. Surprise - AI was everywhere 

AI was everywhere at KubeCon Atlanta. Booths, keynotes, hallway chats, and every corner of the conference were filled with conversations about AI.

Attendees wanted practical answers about real workloads. Most questions were about ML pipelines, infrastructure automation, debugging, and operator intelligence. People wanted to know how to schedule GPU resources, run training jobs efficiently, and confirm that clusters can truly support AI demands.

Reddit captured the mixed mood well. Some joked about nonstop "AI all the things" messaging, while others highlighted real progress. The main challenge was the noise. With so much AI talk, it became harder to tell what each product actually did.

2. Clarity becomes the differentiator

Vendors pushed hard to tell their AI positioning, but the community noticed that the messaging started to sound the same.

One first-time attendee mentioned needing 'a bingo card or drinking game for the amount of managed platform or telemetry services that offered the same exact unique features.' Another noted that after talking with the first five platform vendors, 'it all blended together.' These weren't mean-spirited observations, but they reflect genuine difficulty in understanding what made each solution distinct.

The takeaway is that people want straightforward explanations. They want to know what a product does, how it fits with their tools, and what problems it solves. Clarity helps everyone and leads to better products.

3. What resonated at our booth

Test visibility across distributed pipelines

Teams consistently described similar pain points. Tests scattered across multiple pipelines. Several CI/CD systems running simultaneously - Jenkins (the exodus from it continues…), GitHub or GitLab, ArgoCD, etc. Fragmented tooling that makes it hard to get a complete picture. No single source of truth for test results, history, or trends.

When we showed how Testkube centralizes test execution and results, people got it right away. Having one place to manage all your testing needs, no matter where they come from or what tool is being used, resonated. This isn't about selling a feature. It's about solving a problem every platform team knows.

Testing doesn't have to stay inside CI tooling  

Better still, T (for testing) is conspicuously missing in CI/CD, don’t you think? Many visitors said they hadn't thought about running tests outside of commit-triggered CI. Most teams assume testing only happens after code changes. Someone told us Testkube is to testing what Jenkins is to continuous integration, or Argo to deployments. 

But what about scheduled infrastructure tests? What about running the same E2E tests in a local dev environment? What about checking cluster health before a big deployment? What about testing SLOs and service reliability all the time? When we talked about using Kubernetes clusters as test environments, on schedules or triggered by events, people got intrigued by the opportunities this creates for shifting both left and right. Testing started to feel more flexible and more integrated with real operations.

Real demand for automated infrastructure testing

KubeCon talks showed how dynamic Kubernetes infrastructure is getting. DRA for GPU management, advanced scheduling, and topology-aware placement are changing what infrastructure validation needs to be.

Teams need ways to validate their clusters before running expensive training jobs or rolling out critical releases. They want to test that resource allocation works correctly, that network policies are properly configured, and that the infrastructure layer itself is ready before it is handed over to application workloads. This matches a rising interest we saw in Testkube's ability to run infrastructure tests, not just application tests.

4. Testing is a missing piece

Testing vendors did not have the same presence as platform engineering, security, or AI, but the interest was unmistakably there. Several attendees told us they were genuinely glad to see a testing-focused company on the floor. 

People wanted to talk about continuous testing, environment validation, cluster readiness, and infrastructure testing. As Kubernetes grows more complex, the need for testing is growing just as quickly.

We are helping to move testing from an afterthought to a first-class concern, and KubeCon Atlanta made it clear that teams are ready for this evolution.

5. We got to jam

Some of the best parts of KubeCon happened outside the scheduled sessions, and our team felt that all week. With the wifi down, it almost encouraged even more face-to-face conversations!

The Testkube customer dinner brought together practitioners we’d only met remotely, and we finally connected for IRL conversations. The energy around KubeJam, our community music event, showed how much creativity and passion exist in this space beyond the technical work.

People even stopped by the booth just to say hello after recognizing us online (thank you!). These moments of connection and shared experience are what make KubeCon special

What this means for 2026

From vendor conversations and community feedback at KubeCon Atlanta, three signals stand out for where the cloud native ecosystem is going:

Clarity beats hype

Teams want practical tools with clear value. Practitioners expect vendors to explain what problems they solve and how they fit into existing workflows. This raises the quality of technical communication for everyone.

Testing is moving closer to the center of platform engineering

Kubernetes complexity is forcing teams to rethink testing architecture, not just tools. Good unit tests in CI aren't enough. Platform teams need strategies for continuous testing, infrastructure validation, and operational quality. Testing is now a core platform capability, not just a developer concern.

Infrastructure testing is growing

As Kubernetes takes on more complex workloads like AI training jobs, DRA-managed resources, and topology-aware scheduling, the need for infrastructure testing grows. Teams can't afford to find issues during production. Validation needs to be continuous and systematic. Expect more investment here in 2026.

KubeCon Atlanta 2025 showed a community that's curious, growing fast, and ready to mature its approach to testing and quality. The conversations were honest, the enthusiasm was real, and the path forward is clearer than ever.

Like what you hear? Explore more! 

If these themes resonate with you, we'd love to continue the conversation. Here's how to connect with Testkube:

• Join our community Slack to talk more about KubeCon, testing strategies, and cloud native trends.
• Explore the Testkube sandbox if you want to see what unified testing in Kubernetes actually looks like.
• Watch for details about our next KubeJam in Amsterdam where we'll bring the community together through music once again.

Thank you to everyone who stopped by our booth, attended KubeJam, joined us for dinner, or had great hallway conversations. See you at the next KubeCon!