Key Points
Elon Musk has been a significant catalyst for Nvidia (NASDAQ: NVDA) shareholders recently, and the momentum continued on Friday (Sept. 25). Via a post on his “X” platform, Musk provided an update on Colossus 2, the SpaceX (NASDAQ: SPCX) xAI unit’s AI computing cluster. He stated that the massive training supercomputer could potentially double its deployed Nvidia graphics processing units (GPUs) by year-end.
Musk specified that Colossus 2 currently houses 110,000 Nvidia GB200 chips and 440,000 GB300s. An additional 220,000 GB300 chips are scheduled to activate next week, followed by another 220,000 GPUs in November. He expressed hope for an additional 220,000 GB300s in late December, “if we get lucky.”
SpaceX initially aimed to scale to 1 million GPUs by the end of 2026, a target it appears poised to exceed if these orders are fulfilled. Reports indicate xAI spent approximately $18 billion on its initial 550,000 GPUs, and this new order could be even more valuable given the inclusion of the newer GB300 units.
Image source: Getty Images.
Deepening Nvidia and SpaceX Alliance
This update follows Musk’s earlier comments this month stating that SpaceX would exclusively adopt Nvidia chips going forward. The Tesla and SpaceX CEO praised Nvidia’s Vera Rubin platform as the best architecture available and emphasized the value of their partnership. At the time, Musk noted SpaceX would conclude the year with roughly 2 gigawatts of computing power, aiming for closer to 10 gigawatts by 2027 rather than 5. One gigawatt of compute power costs approximately $50 billion to $60 billion, with chips and other Nvidia components comprising about $35 billion of that expense.
Beyond expanding xAI’s computing capacity, Nvidia and SpaceX also announced a collaboration to develop GPUs for the company’s planned orbital data centers. Cosmic radiation can disrupt current chips by flipping bits from 0 to 1 or vice versa, causing significant computational errors. Without a redundant system for verification, these errors are difficult to detect and impossible to reverse. Musk posted on X that the two companies have designed a Vera Rubin NVL72 system optimized for space, with plans to launch it into orbit by late 2027 and begin scaling in 2028.
Nvidia Stock Outlook
SpaceX represents one of the largest ambitions in the AI sector, making its status as an exclusive anchor customer a major win for Nvidia. Naturally, Nvidia’s advantages extend far beyond SpaceX and its relationship with Musk.
Despite rising competition, particularly in the inference market, Nvidia maintains its dominance in AI infrastructure. Its GPUs remain the primary chips for AI model training, and its CUDA software platform continues to provide a significant competitive advantage. Furthermore, the company’s acquisition of “Groq” and its language processing unit (LPU) technology, integrated into its CUDA ecosystem, offers a differentiated approach to the rapidly growing inference market.
Its ARM-based Vera CPUs are also gaining traction in agentic AI. In August, Nvidia announced that SpaceX will utilize its Vera central processing units (CPUs) for agentic AI workloads. Although the CPU market is much smaller than the GPU market, this represents another solid achievement.
Nvidia has evolved beyond a pure GPU company into a comprehensive AI infrastructure provider, further bolstered by software acquisitions like SchedMD and Hugging Face (pending). Trading at a forward price-to-earnings ratio of just 14 times fiscal 2028 (ending January 2028), Nvidia presents a compelling buy case.
Geoffrey Seiler has no position in the stocks mentioned. The Motley Fool has positions in and recommends Arm Holdings, Nvidia, and Tesla. The Motley Fool has a disclosure policy.
