Tesla(US:TSLA) 3 6 Ke·2025-09-11 08:28Group 1 - The core point of the article is the announcement of the termination of Tesla's "Dojo Supercomputer Project," which was once considered a significant initiative in AI training but ultimately failed after five years of operation [1][3] - Elon Musk had previously expressed confidence in the project, claiming it would achieve "computing supremacy" by 2026, highlighting the abruptness of the project's closure [3][5] - The project was led by Peter Bannon, a key figure recruited from Apple, who has since left to start a new venture [1][3] Group 2 - Tesla has raised approximately $17.6 billion and generated over $60 billion in total revenue, with a peak market value of $1.3 trillion [3][5] - SpaceX, another Musk venture, has an estimated investment of $20 billion to $30 billion and is projected to generate over $10 billion in revenue from subscription services by 2025 [5][6] - Despite high revenues, both Tesla and SpaceX have faced challenges in profitability, with Tesla's net profit significantly lower than that of competitors like TSMC and BYD [5][6] Group 3 - The Dojo supercomputer was designed to train Tesla's Full Self-Driving (FSD) system, utilizing a high-performance D1 chip that offers 3 to 4 times the computing power of NVIDIA's A100 chip [20][22] - The Dojo system was intended to achieve an astonishing computing capability of 100 Exa-FLOPS, but faced significant delays and compatibility issues due to its unique architecture [20][22] - Musk's decision to halt the project reflects a strategic pivot, as he has begun procuring NVIDIA H100 chips and collaborating with TSMC and Samsung for alternative AI solutions [24][25] Group 4 - The failure of the Dojo project is attributed to high costs and a significant gap between investment and returns, compounded by Musk's expansive business ambitions [25][27] - Musk's focus on cost management has led to innovative strategies in battery production and rocket manufacturing, but the same approach has created challenges in the AI sector [27][30] - The competitive landscape in AI has intensified, with Musk acknowledging the need for alternative solutions as Tesla has shifted 70% of its training tasks back to cloud platforms like AWS and Google Cloud [41][43]