Nvidia today announced OEMS like Dell Technologies, Hewlett Packard Enterprise (HPE) and Lenovo are building artificial intelligence (AI)-optimized servers that support the new VMware Private AI Foundation with Nvidia. This move is aimed at helping enterprises customize and optimize deployment of generative AI applications based on their unique data.

VMware and Nvidia built upon their 10-year partnership to develop the private AI foundation platform, where organizations can customize models and run AI applications like chatbots, assistants, search and summarizations. Built on VMware Cloud Foundation, the platform leverages Nvidia's accelerated computing architecture to prepare enterprises for successful integration of AI tools.

"We're reinventing enterprise computing after a quarter of a century in order to transition to the future – to accelerated computing and generative AI," Nvidia CEO Jensen Huang said during a VMware Explore 2023 keynote. "Our expanded collaboration with VMware will offer hundreds of thousands of customers – across financial services, healthcare, manufacturing and more – the full-stack software and computing they need to unlock the potential of generative AI using custom applications built with their own data," he said.

The VMware Private AI Foundation platform is expected to improve privacy, choice, performance, data center scaling and cost. It's also designed to accelerate storage, networking, deployment and time to value for enterprises. The platform is set for release early next year.

Touting Nvidia's leadership in accelerated computing, Nvidia VP of Enterprise Computing Justin Boitano told reporters that virtually every OEM in the world is building systems right now that are designed to run the partners' AI platform and improve Nvidia GPU performance. "Private AI Foundation running on these servers bring another great performance leap in the L40S GPU," which is "the most powerful universal data center GPU" in Nvidia's current lineup, Botiano said.

"This is going to be used to supercharge these generative AI applications," he added, citing 1.2-times more genAI performance on the inference side and up to 1.7-times the training performance compared to the Nvidia's previous GPU generation. These systems will be configured with either Nvidia's BlueField-3 data processing units (DPUs) or ConnectX-7 smartNICs to provide the interconnection needed to boost communications across the data center.

"There's a lot of work that we've done to fully optimize performance between the storage and networking and compute to drive as much efficiency through these processors as possible," Boitano said.

So far, Dell, HPE and Lenovo have committed to delivering this full-stack platform with Nvidia and VMware. The VMware private AI foundation with Nvidia will be available as a single SKU product from VMware, but it will also be available directly from Dell, HPE and Lenovo in pre-integrated systems.

This launch represents a significant step in the adoption and success of AI technologies. Enterprises need to customize AI models against their proprietary information before they can deliver business value to internal teams. "That's when the true business value of generative AI is unlocked," VMware Cloud Platform VP Paul Turner said.

Business don't want to turn over customer records, IT tickets or security configurations to a public AI model because "that's basically taking your proprietary data and encoding it into this publicly available thing. That's why the concept of private AI is so important," Turner said.

Although Nvidia wouldn't share how much this technology will cost its customers, Boitano said the price will be based on GPU consumption. "As [customers] scale their environment based on the number of GPUs, the private AI foundation pricing will be relative to that value that they're getting," he said.