Reports from KubeCon North America 2025 highlight a clear theme: AI is reviving and reshaping platform engineering, not replacing it. The Kubernetes ecosystem has exploded from a handful of projects to a dense landscape of tools, and the arrival of strong open-source LLMs (such as the widely discussed DeepSeek moment) reinforces that there will never be a single “universal” platform. Instead, organisations are rediscovering the need to treat internal platforms as products with real users – their developers.

Speakers at the event emphasised that good platforms hide complexity without pretending it doesn’t exist. Developers want simple abstractions when they’re shipping features on “day one,” but operators still need precise control on “day two” when debugging, scaling, or patching real systems. New open-source work like Formæ, an IaC abstraction layer that unifies infrastructure changes into stateful elements, reflects this direction: let developers build declaratively, while giving operators a safe way to make targeted changes without fighting Terraform in every incident.

Hiring trends echo this shift. Companies such as LinkedIn report that engineers who actively contribute to Kubernetes and related open-source projects tend to be productive much faster when they join platform teams. Open-source experience turns out to be a strong proxy for the mindset and skills needed to evolve complex, AI-augmented platforms. For teams building or refactoring internal platforms today, the message from KubeCon is straightforward: invest in platform engineering as a discipline, not just in a pile of tools, and expect AI to amplify that practice rather than eliminate it.