VMware targets 'three core AI cost drivers' with new Private AI Cloud service

Broadcom has unveiled VMware Private AI Cloud as the firm looks to provide more “secure, scalable, and flexible” private cloud capabilities for enterprises.

Announced at the annual VMware Explore conference in Las Vegas, the new private cloud service aims to consolidate cloud and AI infrastructure, according to Ram Velaga, president of Broadcom’s Infrastructure Software Group.

“VMware Private AI Cloud is the inflection point where enterprise private cloud and private AI infrastructure stop operating as separate disciplines and become one—enabling production inference workloads and agentic AI with the data sovereignty, compliance posture, and cost predictability their business demands,” Velaga commented..

The move comes amidst growing concerns over AI-related costs in recent months, with enterprises contending with surging infrastructure costs and rising AI spending.

According to Broadcom, the new Private AI Cloud service will target “three core AI cost drivers” for enterprises. These include hardware CapEx, operational complexity, and token economics-related challenges.

VMware Cloud Foundation (VCF) 9 will act as a core component within the new Private AI Cloud service, according to Broadcom. New features for the service, which launched in August 2024, aim to lower hardware costs through NVMe memory tiering, as well as cluster-wide storage deduplication capabilities.

New VCF features will also allow enterprises to better-optimize token and infrastructure resource usage. This includes token monitoring tools and enhanced GPU/vGPU tracking.

VMware AI Factory gains

In addition to VCF updates, the new VMware AI Factory will also play a key role in streamlining private cloud and AI innovation, according to Broadcom.

This will serve as the software layer for Private AI Cloud, providing tools aimed at streamlining AI production and deployment timelines. Key features include new multi-tenant model sharing capabilities to reduce infrastructure strain and compute wastage.

Elsewhere, the new AI Gateway will offer users “unified model governance” spanning both on-premises and cloud environments through a single interface.

Broadcom said this will enable access to “locally hosted models with enhancements like intelligent prompt routing, token and usage rate-limiting, and application authorization”.

“We give customers a software-defined foundation that automates infrastructure deployment, unifies lifecycle management, and lets them choose their preferred hardware and vetted models,” said Paul Turner, CPO for the VMware Cloud Foundation Division.

“The result is faster time to first model, predictable private cloud costs, and better control over AI tokenomics.”

Models as a service

VMware customers will also gain access to a wider range of AI models to help provide a “production-ready path” for on-premises adoption.

More than 150 open source and commercial models are available through Private AI Cloud, including Nvidia’s Nemotron 3, Google Gemma 4, Qwen, and GLM 5.2.

Findings from Broadcom’s 2026 Private Cloud Outlook report show more than half (56%) of enterprises are already running - or plan to run - AI inferencing on private cloud.

Improving ease of use on this front is key, according to Chris Wolf, global head of AI and advanced services for the VMware Cloud Foundation Division.

“Broadcom is committed to giving enterprises the broadest set of AI models for their on-premises infrastructure, all validated on VCF,” Wolf said in a blog post.

“Working with the world’s leading AI model providers, we’re giving organizations a clear path to data sovereignty and cost-effective AI at scale, with leading models available securely and delivered as a service to their user community through VMware Cloud Foundation’s built-in services.”

FOLLOW US ON SOCIAL MEDIA

Follow ITPro on Google News and add us as a preferred source to keep tabs on all our latest news, analysis, views, and reviews.

You can also follow ITPro on LinkedIn, X, Facebook, and BlueSky.