Dell unveils AI partnership with Nvidia, aims for on-premises deployment
Dell Technologies and Nvidia have jointly announced the Project Helix initiative to help businesses create and deploy their own on-premise generative AI models.
Both companies will work together to provide “a series of solutions” aimed at enterprises to help use their proprietary data to enhance the value of AI models, all while protecting their data privacy.
Assistance provided through the new offering will span the complete AI journey, from modelling to deployment and fine-tuning results, with Dell providing optimised hardware and software to support this at every stage.
Nvidia’s tested AI Enterprise software will come included with Project Helix, including its more than 100 development tools, frameworks, and pre-trained models.
“This is a full-stack solution that enables enterprises to create and run custom AI models built with their knowledge of their business,” said Chuck Whitten, co-COO at Dell Technologies.
“Our powerful infrastructure and software, Nvidia’s accelerators, AI software, and expertise; this is unmatched, second to none in the industry.”
Whitten pointed to existing Dell hardware that can tackle heavy generative AI workloads such as the PowerEdge XE9680 server.
This contains eight of Nvidia’s H100 GPUs which comprise the core infrastructure on top of which customers can make use of Dell’s server and storage software.
The pair will work to promote on-prem AI, in contrast to the current dominance of public cloud and hyperscaler dominance in AI.
In recent months, Microsoft and Google have established market leadership with their respective chatbots and AI productivity tools.
“Enterprises are working to deploy these models and workloads on-prem to make full use of their data,” said Jensen Huang, CEO at Nvidia.
“If you think about how they drive better insight and better outcomes from their business, they’re looking at ‘I need the context of my business with this data, with these models and tools to really drive faster insights to my business’. They have to do it on-prem because that’s where their data is.
“And they have to do it close to the edge because that’s the speed of light, and you want it to respond instantaneously. But you also want it to be at the edge because, in the future, you want to have information from multiple modalities.
“The more contextual information we get, the better grounded and better inference we can make.”
The announcement was made during the day-two keynote at Dell Technologies World 2023 in Las Vegas, after having been heavily teased by Dell CEO Michael Dell throughout the events of the first day.
Flexibility has been a core tenet throughout the conference, with cloud choice having been a key factor in Dell Technologies’ massive Apex expansion, and this approach applies to Project Helix too.
“There will be a lot of flexibility for customers that are looking to figure out what models to use,” said Varun Chhabra, SVP ISG and telecom marketing at Dell Technologies.
“Helix will come with foundational models that are provided and developed by Nvidia, if customers want to use that. Customers will also have the choice flexibility to be able to pick an open source model of their own choice.
“Once these foundational models have been identified by the customer Helix will help them tune that model with their own proprietary, high-value data sets.”
Chhabra noted that the majority of customers will start with a foundational model, whether this is one included as part of Helix or an open-source model, but that the initiative will provide the flexibility for either to be easily trained on a business’ data.
Dell has stated that its validated designs based on Project Helix will become available in July 2023.
Ⓒ Future Publishing
Subscribers 0
Fans 0
Followers 0
Followers