The cloud-native community that turned Kubernetes into an enterprise standard is now being asked to carry artificial intelligence into production. As the underlying technology looks more alike from one company to the next, the hunt for domain-specific AI value is spotlighting how AI is put to work – and how to do so safely.

That shift is on display this week in Salt Lake City, where the Cloud Native Computing Foundation has added a dedicated AI inference and agentic track to its flagship conference. The plumbing built for microservices is proving to be the substrate for AI factories and agents, along with a renewed focus on tried-and-tested security and business principles, according to John Furrier (pictured, left), co-founder and executive analyst of theCUBE Research.

“The AI-native world that we’ve been covering and following and living — AI factories, the rise of agentic, which involves coding and now tasks — are all based fundamentally on the substrate of cloud-native technologies,” Furrier said. “This community is the heart and soul for the AI generation.”

Furrier spoke with theCUBE’s Rebecca Knight (right) as a part of a keynote analysis at KubeCon + CloudNativeCon NA, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed how enterprises can secure and govern AI agents while turning proprietary knowledge and workflows into measurable business value rather than simply maximizing token consumption. (* Disclosure below.)

Why domain-specific AI value beats tokenmaxing

Kubernetes has matured from a niche orchestration project into the reference platform for major clouds and neoclouds, and enterprises are now standardizing on it. That maturity is what makes it useful for AI, which needs the same control planes, scalable data layers and telemetry the community already worked out for microservices, according to Furrier.

“Kubernetes got boring, which means it’s working,” Furrier said. “A lot of things that are happening in cloud-native that were involved in microservices in Kubernetes world are actually the same challenges in the AI world.”

Those lessons now apply to a pressing question: how much freedom enterprises will give AI agents inside real systems. Agents that cause trouble are usually poorly instructed rather than malicious, and the answer lies in context, identity and guardrails, according to Furrier. Perimeter-style security thinking is expected to make a comeback, along with a move away from token consumption as a measure of value and toward tangible business outcomes.

“We’re gonna get back to an era of, ‘Build something, show me what you built, present it for value,'” Furrier said. “If it’s worth something, there’s an exchange for value.”

Stay tuned for the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of KubeCon + CloudNativeCon NA.

(* Disclosure: TheCUBE is a paid media partner for the KubeCon + CloudNativeCon NA 2026 event. Sponsors of theCUBE’s event coverage do not have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE