Elon Musk has instructed Nvidia to reroute AI chip deliveries to his social media firm X (formerly Twitter) and AI startup xAI, causing delays for Tesla. Internal emails accessed by CNBC reveal that this decision has pushed back Tesla's receipt of more than $500 million worth of chips by several months.
Impact on Tesla's AI Development
According to Nvidia's internal communications, Musk directed the company to prioritize AI chip deliveries for X, rather than Tesla. Nvidia emails reveal that Musk instructed the shipment of thousands of GPUs originally set aside for Tesla to X and xAI. Staff noted that Musk's statements during Tesla's first-quarter earnings call on GPU investments conflicted with actual orders. xAI reportedly rents around 16,000 Nvidia H100 GPUs from Oracle Cloud and uses AWS, with some capacity at X data centers, according to CNBC.
Consequently, this has delayed essential hardware that Tesla needs for its AI and robotics projects like the Tesla Bot. On the recent earnings call, Musk had projected an increase in Tesla's Nvidia H100 AI chips from 35,000 to 85,000 by year-end. However, Nvidia staff indicated that Musk's public claims about Tesla's AI expenditures and procurement might have been overstated.
According to CNCB, an Nvidia memo from December mentioned Musk diverting H100 GPUs initially meant for Tesla to X. Another document, post-earnings call, pointed out inconsistencies with the $10 billion AI expenditure Musk announced relative to Nvidia's forecasts and observing job cuts at Tesla.
Tesla will spend around $10B this year in combined training and inference AI, the latter being primarily in car.
Any company not spending at this level, and doing so efficiently, cannot compete.
— Elon Musk (@elonmusk) April 28, 2024
Musk has asserted that AI capabilities are central to Tesla's future, requiring substantial GPU resources. These decisions may slow Tesla's progress but might temporarily aid financial stability amid struggling profits.
Nvidia did not provide a comment, and Musk did not respond to requests for remarks. Following CNBC's report, Musk clarified on X about Tesla's AI-related spending plans, including investments in the Tesla-designed AI inference computer and sensors, alongside Nvidia hardware costs. He estimates Nvidia purchases by Tesla this year to be between $3 billion and $4 billion.
Of the roughly $10B in AI-related expenditures I said Tesla would make this year, about half is internal, primarily the Tesla-designed AI inference computer and sensors present in all of our cars, plus Dojo.
For building the AI training superclusters, NVidia hardware is about…
— Elon Musk (@elonmusk) June 4, 2024
Musk Says Tesla Cannot Use the Nvidia Chips Immediately
Musk mentioned that Tesla lacked a facility to utilize the Nvidia chips immediately and that they would have remained in storage. He added that Giga Texas' southern extension, which would house 50,000 H100s for FSD training, is almost complete.
Tesla had no place to send the Nvidia chips to turn them on, so they would have just sat in a warehouse.
The south extension of Giga Texas is almost complete. This will house 50k H100s for FSD training.
— Elon Musk (@elonmusk) June 4, 2024
Musk also stated that xAI plans to deploy a substantial GPU supercomputer by next summer and will soon launch a data center with 100,000 H100 GPUs. These claims remain unverified.
Conflicts of Interest and Shareholder Concerns
Musk's directive to prioritize X and xAI has stirred concerns of conflicts of interest, given his executive roles across multiple companies. Legal experts suggest that this allocation might divert opportunities from Tesla to Musk's other ventures.
Tesla's shareholders are increasingly questioning whether Musk can adequately juggle his responsibilities between Tesla and his other enterprises such as SpaceX and Neuralink. The postponement of chip deliveries exacerbates these concerns, especially as Tesla battles growing competition and declining sales.
Musk's actions have faced criticism from Tesla's largest retail shareholder and other investors. They argue that demanding more control over Tesla's operations could be problematic. The rerouting of AI chips from Tesla to X seems to reflect Musk's tendency to blend resources across his ventures, potentially harming Tesla's AI advancements and business performance.
Nvidia's Allocation Challenges
Nvidia, one of the most valuable companies globally, is facing intense demand for its AI chips from industry giants like Google, Amazon, Meta, and Microsoft. CEO Jensen Huang admitted the challenge of meeting this soaring demand, noting that customers are “consuming every GPU that's out there“. Nevertheless, Musk has secured a large allocation of Nvidia's AI chips for X and xAI, while Tesla shareholders are concerned about the implications for the automaker's AI projects.
