LAS VEGAS–At it’s annual Explore event, VMware by Broadcom made the case for bringing (back) IT workloads from the public cloud to on premise data centers. In addition to addressing the cost and complexity of public cloud, the pitch also covered the idea of private AI—essentially bringing AI capabilities into local private clouds to bolster performance and compliance while delivering on high value use cases. The company provided updates on its work with NVIDIA in the private AI space initially announced at Explore last year.
VMware Private AI Foundation with NVIDIA is billed as “a comprehensive solution for deploying, managing, and scaling AI-driven applications securely and efficiently on [VMware Cloud Foundation (VCF)]-based private clouds.” A wide range of features are available in the current version of VCF, 5.2.1, with more enhancements coming in the next major release, VCF 9. More on that here.
The solution incorporates NVIDIA Inference Microservices (NIM), a set of containerized microservices designed to simplify and accelerate AI model deployment, and the NVIDIA AI Enterprise software platform which includes a broad set of AI tools optimized for NVIDIA GPUs.
When it comes to private AI, “We’re not messing around”
In a session with media and analysts, Chris Wolf, global head of AI and advanced services in Broadcom’s VMware Cloud Foundation Division, laid out common private AI use cases, including code generation to accelerate development, contact center resolution, IT operations automation, and advanced information retrieval. For the contact center resolution use case, Wolf said private AI makes sense because it’s using a companies’ private data; and running it on prem, he said, delivers material cost savings.
In addition to cost reduction, Wolf said private AI bolsters policy and control, allows for resource sharing, and helps centralize operations. “From a market performance perspective, this is the fastest growing new product I’ve ever been associated with.” He called out early traction with financial services and public sector customers, as well as in manufacturing, oil and gas, and retail.
“The fact that we can automate these applications and AI services end to end and deliver them in minutes is a real core strength of ours,” Wolf said.
An interesting capability coming with the release of VCF 9 is around GPU virtualization wherein a physical GPU can be shared by multiple virtual machines; this lets the enterprise maximize GPU utilization which would equate to faster return on what’s likely a significant investment. Also on deck is a tool for virtualized GPU profile visibility so administrators have an easy interface to view all virtualized GPUs across their GPU footprint. And be on the lookout for GPU reservations to improve capacity planning and operational efficiency. Read more in this blog post.
Summing things up, Wolf said the VMware Private AI Foundation with NVIDIA represents a full suite of “industry leading capabilities…We’re not messing around. We hear you. You’re all saying, ‘Broadcom, VMware, you need to do more with AI.’ We are…I really care about this stuff and, as a technologist, I really think we’re on the right path.”
