Tesla's Nvidia AI Chips Reallocated by Musk, CEO Explains

Jun 05, 2024 01:29 AM EDT

RED TESLA CAR — (Photo : AI IMAGE/DABBIE DAVIS)

Elon Musk reportedly directed the transfer of numerous Nvidia-manufactured AI chips from Tesla to his social media venture X. Musk cited Tesla's current insufficiency in storage and activation capacity for the Nvidia processors. Internal emails within Nvidia discussed this development.

Nvidia Transferred AI Processors for Tesla to X

Elon Musk, the CEO of Tesla, has directed Nvidia to prioritize the delivery of AI processors to his companies X and xAI over Tesla. According to Bloomberg, Elon Musk verified redirecting AI chips from Tesla Inc. to X Corp. and xAI Corp., providing rationales for the switch and internal Nvidia Corp. emails questioning Tesla's procurement strategies.

Musk stated on X that Tesla lacked suitable space to activate Nvidia chips. CNBC previously reported Musk reallocating 12,000 H100 graphics processing units from Tesla to X based on a December Nvidia memo.

In a late April email seen by CNBC, Nvidia employees noted discrepancies between Musk's remarks on Tesla's earnings call and X regarding the company's commitments with Nvidia. Musk mentioned during the April 23 earnings call that about 35,000 H100s were in use, projecting around 85,000 operational units by year-end.

Additionally, on April 28, Musk mentioned on X that Tesla plans to invest approximately $10 billion on AI this year. Fox Business highlighted that, as reported by CNBC, the modification postponed Tesla's acquisition of over $500 million worth of processors for several months.

Furthermore, Fox Business reported that it was mentioned that Tesla would soon integrate Nvidia's AI chips for training its Full-Self Driving (FSD) software at the Gigafactory located in Austin, Texas. The statement indicated that the south extension of Giga Texas is almost complete. This will accommodate 50k H100s for FSD training.

Following this, Musk mentioned that roughly half of the $10 billion earmarked by Tesla for AI expenditures will be utilized internally, with Nvidia acquisitions anticipated to total between $3 billion and $4 billion.

He emphasized, the challenge of coordinating 50K H100s for training as a unified system cannot be overstated. No company globally has successfully accomplished this task to date. Once they establish the functionality of this system, further hardware purchases will ensue.

According to Autoevolution, the recent developments have led Tesla investors to question Musk's ability to fulfill his responsibilities. Musk and Tesla are currently entangled in a string of legal battles with investors alleging market manipulation and insider trading. The ongoing sales downturn and challenges on the horizon for Tesla only add to the concerns.

Moreover, as reported by Autoevolution, the Nvidia correspondence indicates Musk's inclination to give precedence to GPU deliveries for his alternative enterprises. In addition to the chips, Musk has transferred personnel from Tesla, such as machine-learning expert Ethan Knight. Furthermore, at least four other ex-Tesla staff members engaged in Autopilot and big data initiatives at Tesla have been reassigned to x.AI.

Nonetheless, in a statement on X, Musk refuted any notion that redirecting the Nvidia chips to X and x.AI posed a disadvantage to Tesla. In contrast, he asserted that this choice guaranteed timely utilization of the chips, especially given the unfinished state of Tesla's Giga Texas data center.

Musk highlighted that had the chips been sent to Tesla, they would have remained idle in a warehouse.

About Nvidia

According to its website, NVIDIA was the trailblazer in accelerated computing, addressing challenges that are beyond the reach of others. The company's endeavors in AI and digital twins are reshaping major global sectors and significantly influencing society.

Moreover, as stated in a document from its website, the extensive NVIDIA ecosystem encompasses 4.5 million developers, 40,000 companies, and more than 3,300 applications.