Agentic AI infrastructure shifts enterprise focus from model choice to platform control
As agentic AI infrastructure moves from experimentation into production, enterprises are confronting a more complex question than which model to use: how to control the cost, data exposure and infrastructure supporting production AI applications.
That shift is pushing organizations to rethink how much they should rely on public cloud AI services alone, especially as agentic systems move from simple assistants into persistent enterprise applications that act across business systems. The turning point comes when companies begin treating AI not as a pilot project but as an operating model with enterprise-scale consequences, according to Joe Fernandes, vice president and general manager of the Artificial Intelligence Business Unit at Red Hat Inc.
“The token costs … are exploding rapidly, particularly as you move from experimentation to at-scale production systems and from simple chatbots and assistants into these always-running enterprise agents,” Fernandes said. “I think cost is a huge factor, but then there’s also the data side: Are you willing to let your data go into these public cloud services? Do you have compliance or sovereign requirements that preclude that? I think those are the two things combined … that make them start thinking, ‘Maybe there’s an alternative here to just exclusively relying on these public cloud services.’”
Fernandes spoke with theCUBE’s Rob Strechay during an exclusive CUBE Conversation about the rise of agentic AI infrastructure and the expanding role of platform teams. The discussion also examined why enterprises are looking to hybrid, open-source approaches as they build AI systems for production. ( Disclosure below.)*
Agentic AI infrastructure raises the stakes for platform teams
As agents become a new class of enterprise application, the platform function moves closer to the center of AI strategy. Reliability, scale and security still matter, but autonomy adds another layer of operational responsibility, according to Fernandes.
“I think that platform teams are more important than ever because this is the next layer of the enterprise platform, and agents are the next evolution of enterprise applications,” Fernandes said. “But they need to run someplace, and you need a platform. You still need reliability in that platform; you need uptime … scale and security, all the same things that platform teams have needed to focus on for years in their applications. It’s no different as it relates to agents, but there are new things because these are autonomous systems.”
Those requirements grow more complex when enterprises move from a handful of AI assistants to hundreds or thousands of agents acting across systems. That’s where hidden operational costs emerge — not just in compute, but in controlling what agents can reach and how their actions are traced, Fernandes added.
“What makes an agent an agent is the fact that it’s autonomous,” he said. “What’s it allowed to access in your network, on your file system? So, we introduced the concept of agent sandboxes.”
Red Hat is extending its open-source approach beyond model access to infrastructure and sandboxing required to run agents. That work includes Nvidia Corp.’s OpenShell, an open-source sandbox runtime for AI agents that Red Hat actively contributes to and maintains.
“OpenShell is a key project for us now in the agent sandbox space,” Fernandes said. “We’re always looking for where can we … collaborate with other vendors and other users to try to drive innovation faster and try to drive things toward a standard.”
That same need for control extends to sovereignty and hybrid deployment, where enterprises increasingly want control over where models, agents and data run. Red Hat’s view is that agentic AI infrastructure will have to span public cloud services, private environments, sovereign clouds and the edge, according to Fernandes.
“A lot of governments and organizations need to manage [sovereignty requirements] by being able to control where stuff runs and where their data lives, where those models run [and] where the agents run,” he said. “This just builds on what we’ve been saying for years: This is going to be a hybrid world.”
Here’s the complete video interview with Joe Fernandes:
( Disclosure: Red Hat sponsored this segment of theCUBE. Neither Red Hat nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)*
Image: ChatGPT/SiliconANGLE Media
Support our mission to keep content open and free by engaging with theCUBE community. Join theCUBE’s Alumni Trust Network, where technology leaders connect, share intelligence and create opportunities.
15M+ viewers of theCUBE videos, powering conversations across AI, cloud, cybersecurity and more** 11.4k+ theCUBE alumni**— Connect with more than 11,400 tech and business leaders shaping the future through a unique trusted-based network.
About SiliconANGLE Media
theCUBE AIand theCUBE SuperStudios — with flagship locations in Silicon Valley and the New York Stock Exchange — SiliconANGLE Media operates at the intersection of media, technology and AI.
Founded by tech visionaries John Furrier and Dave Vellante, SiliconANGLE Media has built a dynamic ecosystem of industry-leading digital media brands that reach 15+ million elite tech professionals. Our new proprietary theCUBE AI Video Cloud is breaking ground in audience interaction, leveraging theCUBEai.com neural network to help technology companies make data-driven decisions and stay at the forefront of industry conversations.