Elon Musk’s xAI Collaborates with Nvidia and Dell to Create Advanced AI Supercomputer

Elon Musk's AI startup xAI is forging partnerships with Dell, Nvidia, and Supermicro to develop what it argues will be the largest supercomputer in the world. Announced in early June, this ambitious project aims to establish a “gigafactory of compute,” providing xAI with dedicated computing infrastructure. This will enhance its training capabilities while reducing its reliance on cloud services from Oracle and various data centers.

The collaboration with Dell was highlighted when CEO Michael Dell shared a photo of server racks on social media, revealing the company's active role in creating this supercomputer’s infrastructure, which he refers to as one of its “AI factories.” Dell's AI factory concept centers on offering specialized infrastructure services, utilizing liquid-cooled servers along with Nvidia's newly released Blackwell GPUs to accommodate demanding AI workloads.

Musk confirmed that Dell will assemble half of the supercomputer’s racks. The remaining half will be built by Supermicro, whose participation Musk acknowledged in a reply on social media, simply stating “SMC,” referring to Super Micro Computer. Although Supermicro has yet to publicly confirm its involvement, it is evident that both companies are integral to the project.

In terms of hardware, Musk disclosed plans to acquire around 300,000 Blackwell B200 GPUs from Nvidia, with the acquisition anticipated for "next summer." He expressed skepticism about investing in Nvidia's current flagship H100 GPUs due to the rapid advancements in AI hardware. This decision follows Nvidia’s emergence as the most valuable company globally, partially attributed to its annual release schedule for new GPUs, including updated versions of the Blackwell line and the new Rubin chips.

xAI has recently raised $6 billion in funding and, within just one year of its inception, boasts a valuation of $24 billion. Notably, Musk's startup has been receiving shipments of Nvidia chips that were initially allocated for Tesla. Reports indicate that approximately 12,000 H100 GPUs, originally intended for Tesla, were redirected to xAI. During the company’s first-quarter earnings call, Musk mentioned Tesla's plan to acquire around 85,000 Nvidia H100 GPUs, but noted that the company had “no place to send the Nvidia chips to turn them on,” leading to their diversion from Tesla to xAI.

By establishing its own supercomputer infrastructure and aligning itself with leading tech companies, xAI is poised to make significant strides in the AI landscape, particularly in the development of its chatbot, Grok, aimed at competing with established models like OpenAI’s ChatGPT.