We’re feeling cynical about xAI’s big deal with Anthropic

xAI, a subsidiary of SpaceX, has finalized a significant deal with Anthropic, offloading the entire compute capacity of its Colossus 1 data center in Tennessee. This partnership has sparked discussion regarding xAI's strategic direction and its implications for SpaceX’s forthcoming initial public offering (IPO). The move suggests a pivot for xAI towards becoming a "neocloud" provider, renting out computing resources rather than primarily focusing on developing its own frontier AI models. This strategy contrasts with many leading AI companies that retain their compute power for internal model training. While the deal offers xAI a new revenue stream, it also casts doubt on its commitment to innovation in AI development. Critics argue that relying on renting out GPUs may hinder xAI's ability to brand itself as a forward-looking AI innovator. This shift in focus is particularly salient given previous reports that xAI employees were not even using their own AI models internally, leading to significant internal restructuring. Furthermore, the partnership emerges amidst reports of xAI’s impending dissolution as a separate entity within SpaceX, with Elon Musk reportedly planning to integrate it more closely under the "SpaceXAI" umbrella. This move, coupled with the sale of its data center capacity, suggests a re-evaluation of xAI's initial vision. The deal with Anthropic, therefore, appears to be a calculated move ahead of SpaceX's IPO, offering a more immediate and "believable business" model to potential investors, even if it tempers long-term excitement about xAI's AI ambitions.