Dell adds Nvidia's next GPUs to its portfolio of AI platforms
Nvidia is a kingmaker, and who wouldn't want to be king?
Dell has tied its AI flag firmly to Nvidia's mast with its latest offerings, comprising a fully integrated end-to-end platform for enterprise customers looking to build and operate their own AI, plus updated server support for Nvidia's upcoming GPUs, scalable storage and professional services.
To be announced at Nvidia's GTC event in San Jose today, Dell's portfolio of AI products pair its hardware with GPUs, networking and AI software from Nvidia in order to meet the expected frenzy of demand for AI support.
Dell is also updating its PowerEdge servers to support GPUs expected to be detailed by Nvidia at GTC, which the acceleration champ hinted at earlier this month. The update primarily revolves around its flagship AI system, the XE9680, which has been designed to cram in up to 8 GPUs in a single chassis.
The system will now support the H200 GPU as well as the upcoming Blackwell-based B100 and B200, covered in a separate Reg article detailing Nvidia's announcements at GTC. Dell also revealed that the XE9680 is being enhanced to include liquid cooling for the first time in order to support the B200 product.
According to Varun Chhabra, Dell's SVP of infrastructure and telecom marketing, Dell is offering "ultimate Nvidia GPU flexibility" with this update, "starting with H100 available today to the H200, or in the future, the B100 and the B200."
The Texan IT giant will also be supporting Nvidia's GB200 Superchip – the successor to the GH200 which will feature a Blackwell GPU instead of the Hopper of the current product – but it isn't clear when this will be available.
This is not the first wave of AI products from Dell. Last year it introduced a validated platform for inferencing, followed by another for model customization and tuning.
Now, it's filling in the final pieces in the AI pipeline with Dell Generative AI Solutions for Model Training and Dell Generative AI Solutions for Retrieval-Augmented Generation (RAG) – the latter of which allows customers to enhance the accuracy of AI models by bringing in their own proprietary data.
- Can AI shorten PC replacement cycles? Dell seems to think so
- Oracle investors hear the magic word 'Nvidia' and boom! Buy, buy, buy
- Dell exec reveals Nvidia has a 1,000-watt GPU in the works
- Incoming wave of AI is making buying PCs riskier for businesses
Confusingly, Dell is also announcing a separate, fully integrated AI solution known as Dell AI Factory with Nvidia, delivered jointly by Nvidia and Dell Services. This comprises a rack-scale design using Dell systems with GPUs, Nvidia's Spectrum-X networking, and the AI Enterprise suite and NeMo framework for large language models (LLMs).
"What we found from enterprises is the most common question is 'give us an easy button – give us a way to deploy AI solutions fast with reduced time to value,' and that's what the Dell AI Factory with Nvidia does," explained Chhabra.
Dell AI Factory with Nvidia is available globally through traditional channels and its Apex IT as-a-service offering. Also available now is the Dell Generative AI Solutions with Nvidia – RAG, while Dell Generative AI Solutions with Nvidia – Model Training will be available in April.
Continuing the Big D's love-in with Nvidia, the Texan outfit declared that its PowerScale network attached storage (NAS) platform is the first Ethernet storage system validated for use with Nvidia's DGX SuperPOD AI infrastructure. The latter is a cluster of custom servers crammed with GPUs for AI processing.
Because the PowerScale platform is a scale-out architecture that can expand to up to 252 nodes, it can "easily exceed all the benchmark requirements for DGX SuperPOD," Dell claimed.
Another announcement concerns Dell Data Lakehouse for AI, which will be available as of March 19. This is the "open, modern data lakehouse" platform Dell disclosed it was developing last year in partnership with Starburst to support a broad set of AI workloads and analytics.
This is a federated solution designed to enable customers to get access to their data much more quickly, Dell claims – minimizing data movement by discovering, querying and processing data in-place, including at the network edge.
"It helps optimize for AI with a complete end-to-end offering all the way up and down the stack that is fully tested and integrated," explained Greg Findlen, Dell's SVP of AI and data management solutions.
Finally, Dell is expanding its professional services for AI with offerings to help customers deploy – including consulting services for the Dell Data Lakehouse. Also coming are Infrastructure Deployment Services for Model Training, available in select locations from March 29, and Implementation Services for RAG, available in select locations from May 31. Advisory Services for GenAI Data Security will also be available in select countries starting March 29.
Earlier this month, a datacenter report from analysts at Omdia claimed that Nvidia's importance to the server market is now so great that it is effectively a "kingmaker", and a close working relationship with it would be vital for any server maker wanting to grow its share of the market this year. This is obviously a lesson that Dell has taken to heart. ®