Hewlett Packard Enterprise has reportedly entered a high-value contract to supply X, Elon Musk’s platform formerly known as Twitter, with AI-enhancing server technology surpassing $1 billion in value.
The deal was initially highlighted by Bloomberg, citing unnamed sources who suggested that HPE would further Musk’s ambitions in artificial intelligence.
Musk’s xAI, known for deploying its Grok model via X, operates a supercomputer named Colossus in Memphis, utilizing Dell and Supermicro systems alongside 100,000 GPUs.
Plans are underway to boost this setup with another 100,000 GPUs, with aspirations to eventually scale to one million GPUs for comprehensive AI processing.
Speculated placements for the HPE systems include X for AI operations or within xAI, given the intertwined nature of both entities, with xAI enabling users to interact with the Grok model via X. Previously, Musk reallocated Nvidia H100 GPUs from Tesla to xAI for efficiency.
Recently, xAI widened access to Grok’s new image-generation features to more users on X, leading to anticipated demand for enhanced computational resources as its models expand.
HPE, leveraging its Cray division’s expertise, continues to hold a prominent position in supercomputing, crafting leading systems like El Capitan, Frontier, and Aurora.
Their latest GPU systems offer a choice of Nvidia, AMD, or Intel components, although xAI’s preference remains undisclosed, many Musk enterprises favor Nvidia.
This strategic selection from HPE might also serve as a diversification pivot following issues encountered by Supermicro.
HPE and X have yet to comment. The Register awaits further details on this pivotal collaboration.®