Tesla is ceasing its own artificial intelligence supercomputer efforts, meaning the company will rely more heavily on major chipmakers such as Nvidia and Advanced Micro Devices (AMD). According to Wells Fargo analysts, this shift will likely increase demand for general-purpose GPUs from these companies. Tesla’s custom-built AI training supercomputer, Dojo, was shut down due to the high costs associated with its development.
Tesla’s hardware design engineering vice president, Pete Bannon, is leaving the company after joining from Apple in 2016. Remaining employees are being reassigned to work on other projects within the company. The move suggests Tesla is stepping back from its internal AI-optimized silicon efforts.
However, this decision could be beneficial for Nvidia and AMD, as it indicates an increase in Tesla’s usage of general-purpose GPUs for AI applications. Tesla has been expanding its graphics processing units (GPUs) infrastructure to scale its training compute power. The company has bought thousands of Nvidia GPUs, including the H100 and H200 models, for this purpose.
In contrast, Elon Musk previously mentioned that Tesla would spend more than $500 million on Nvidia hardware this year, rather than using a custom-built supercomputer. This discrepancy highlights the changing landscape of Tesla’s AI efforts and its reliance on external vendors for GPU capabilities. Despite this shift, Tesla shares have been relatively stable, with a 20% decline in stock performance over the past year.
Source: https://www.cnbc.com/2025/08/08/why-teslas-latest-move-could-be-a-boon-for-nvidia-amd.html