Dell opens a portal on the open source Hugging Face platform to provide companies with the resources necessary to run generative AI models on on-prem servers.

Anyone who wants to work with generative AI today is often forced to work with a hyperscaler. To run generative AI on-premises, you first need the technical knowledge to develop a model, but also the server capacity to house the models. Dell wants to make on-site generative AI a fully-fledged alternative and sees Hugging Face as the right partner.

Hugging Face is an open source application development platform. The company attaches great importance to its members’ sense of community. Anyone can share machine learning models and datasets to train AI with each other. Dell will position itself as a generous benefactor in the collaboration.

Custom containers

Dell offers containers and scripts to implement the open source models via Hugging Face. These containers and scripts are tailored to your own infrastructure. Therefore, it is helpful if you already have a Dell server. Over time, Dell will continue to optimize the containers. “Dell’s GenAI capabilities combined with Hugging Face’s collection of datasets and libraries offer users the freedom of open source GenAI models with the confidence of data security and on-site reliability,” said Jeff Boudreau, Chief Dell Technologies AI Officer.

Dell isn’t the only party trying to free generative AI from the clutches of hyperscalers. The former subsidiary VMware announced collaborations with Intel and IBM on the topic of “private AI” a week ago as part of its Explore event. This is another description for AI models that are not located in the data centers of an external cloud provider, but rather on an internally managed infrastructure.