Private AI is moving into production, and enterprises are demanding more than another collection of AI components. Product marketing is now about turning that complexity into a turnkey system that can get AI running faster.

The bet is that the platform enterprises already use to run virtual machines and containers can also run their models and agents, delivered as a single turnkey system rather than a stack customers assemble themselves, according to Prashanth Shenoy (pictured), chief marketing officer and vice president of the VMware Cloud Foundation Division at Broadcom Inc.

“We have reached a very critical juncture in the world of AI,” Shenoy said. “A lot of our organizations are moving from pilot to production at scale, so there are big concerns around cost, tokenomics, security and privacy concerns of their data. A lot of our organizations are looking towards private cloud as the preferred platform for deploying their production AI workloads.”

Shenoy spoke with theCUBE’s John Furrier at VMware Explore 2026, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed the VCF AI Factory, private AI services and frontier AI security. (* Disclosure below.)

Product marketing moves toward turnkey AI factories

Without a packaged offering, customers stitch together virtualization, storage, Kubernetes services and a model gallery on their own, which creates silos and slows deployment. Broadcom has certified servers from multiple vendors, worked with Advanced Micro Devices Inc. and Nvidia Corp. on the chip layer and has partnered with MetalSoft Cloud Inc. on heterogeneous firmware and hardware bring-up, Shenoy explained. The goal is to bring those components together into a single operational layer rather than leave customers to integrate them themselves.

“All of this is integrated into the VCF Ops Console,” he said. “The operations become easy, and the provisioning of this hardware and management gets reduced from months [or] weeks, to now minutes.”

Model choice matters as much as hardware certification. About 150 models have been tested and optimized to run on VCF, spanning open-source, open-weight and commercial options, alongside an AI gateway that connects local models to more than 40 cloud model providers, Shenoy noted.

“Not every use case that enterprises have requires a frontier LLM model,” he said. “It’s all about purpose, fit, governance and cost, which is very, very critical.”

Security is the other half of the story, with frontier AI a central theme at this year’s VMware Explore event. Broadcom hardened VCF 9.1 from the ground up, moved to monthly patch releases and is arming practitioners with new certifications. All this is in service of Broadcom’s Frontier AI Security Readiness Program, which covers assessment, architecture, implementation and upskilling, Shenoy said. The program is designed to help organizations prepare for threats that are moving at AI speed.

“The volume, the velocity and the type of variety of these AI-driven threats have just exploded,” he said. “Attackers don’t take weeks or months. They take hours or minutes to get into the system.”

Years of product marketing decisions to collapse the VMware portfolio into fewer stock-keeping units now look like preparation for this moment, with capabilities such as memory tiering built in before the server supply crunch arrived. The payoff is that AI does not require a parallel estate, Shenoy noted.

“You don’t need to create another siloed infrastructure,” he said. “The same infrastructure that you’ve tried and tested for running your VMs – for running your containers – can be used to run your agents and AI workloads, with the same unified operations management and the security and data privacy.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of VMware Explore 2026:

(* Disclosure: TheCUBE is a paid media partner for the VMware Explore 2026 event. Neither Broadcom, the sponsor of theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE