NVIDIA is the third member of this collaboration, which brings multiple new turnkey Generative AI and multi-cloud solutions to market.
Brian Connors, Global VP & GM, Software & Business Development at Lenovo
LAS VEGAS – At the recently concluded VMware Explore 2023 event here, Lenovo and VMware announced the availability of the first turnkey solutions from their joint Edge and Cloud Innovation Labs, which were announced last October at the Lenovo Tech World event. The new solutions include two models with NVIDIA GPU-dense platforms designed for the VMware Private AI Foundation with NVIDIA generative AI solution, which was also announced at the event. They also include new Lenovo computing platforms with three server configurations in one, including support for NVIDIA HGX A100 4-GPU systems with NVIDIA NVLink technology and Lenovo Neptune hybrid liquid cooling, as well as four or eight GPU configurations featuring NVIDIA L40S and 80GB NVIDIA H100 Tensor Core GPUs, or NVIDIA H100 NVL servers. Related announcements included Lenovo unveiling their newest Reference Design for Generative AI, which combines NVIDIA-accelerated computing with their ThinkSystem solution.
“These announcements are primarily between Lenovo and VMware, and reflect the joint announcement made at last October’s Tech World,” said Brian Connors, Global VP & GM, Software & Business Development at Lenovo. “As we evolve, you will see other partners come in. These initial elements are around multi-cloud.”
“Our joint labs were focused on hybrid, multi-cloud and edge,” said David Reeves, Global Alliances Director at Lenovo. “As we collaborated more closely on Generative AI, NVIDIA got pulled in.”
Connors said that while Lenovo’s relationship with VMware is much broader than Generative AI, it represents a great opportunity for them both.
“In the past, the only way to leverage Large Language Models [LLM] was to leverage all cloud providers, which was a very different model from where it is now headed,” he stated. “The need for toolchains to allow enterprise customers to generate their own LLM with their own data is just emerging. Having a simpler way to do it is this phase of long, long road, and is very different from a hardware processor standpoint. I continue to tell my team that we are entering a bubble and we need to take advantage of that.”
The most head-turning part of the announcement is the new NVIDIA GPU-dense platforms that are purpose built for AI workloads, including the Lenovo ThinkSystem SR675 V3 and ThinkSystem SR670 V2. They offer up to eight-GPUs in a compact 3U or smaller footprint for the highest performing accelerated workloads.
“These two models are specific to the VMware Private AI Foundation,” Connors said. “The models designed more broadly for agile and multi-cloud purposes are ThinkAgile systems.”
The other Lenovo platforms that will come from the announcement are for Lenovo TruScale Hybrid Cloud with VMware, and will enable intelligent cloud transformation on demand. They include server configurations that support for NVIDIA HGX A100 4-GPU systems with NVIDIA NVLink technology and Lenovo Neptune hybrid liquid cooling, as well as four or eight GPU configurations featuring NVIDIA L40S and 80GB NVIDIA H100 Tensor Core GPUs, or NVIDIA H100 NVL servers.
Connors said that the extension of this partnership with turnkey solutions would create new opportunities for partners.
“They are turnkey solutions, but think of turnkey as Day Zero,” he said. “It’s about getting it up. The services opportunities could be on the installation side, and on – ongoing management for the whole lifecycle. That’s not turnkey. The turnkey aspect is about making these simple to acquire and to deploy.”
David Reeves, Global Alliances Director at Lenovo
“We think that this will make it easier for partners to provide services around the offerings,” Reeves said. “Most partners haven’t been able to drag the VMware Cloud Foundation down to the midmarket. This will be easier for them to sell in a midmarket environment.”
Lenovo is adding the latest NVIDIA Spectrum-X networking technology from NVIDIA to its AI portfolio, using NVIDIA BlueField-3 data processing units (DPUs) and NVIDIA Spectrum-4 switches in its Generative AI reference design.
“Everyone used Bluefield in Project Monterey, but for Generative AI, it is new,” Reeves noted.
Connors concluded by emphasizing that even though Generative AI is new and the new versions are untested, they are seeing strong customer response and expect it to do well.
“We commonly run incubation programs in Lenovo, including the broadest edge program in the industry,” he said. “Is it fully mature yet? Absolutely not, but it’s not speculative either. Demand is very strong for the highest end machines, but it’s still a very early stage. As skills in managing it evolve, you will see this grow significantly.”
