Tesla يُنهي مشروع Dojo، الحاسوب الفائق لتدريب الذكاء الاصطناعي الذي كان يُعدّ حجر الزاوية في رؤية ماسك للقيادة الذاتية الكاملة

Tesla is dismantling its Dojo supercomputer project, a key pillar of Elon Musk’s vision for in-house AI and full self-driving technology, according to Bloomberg, citing anonymous sources. The move marks a significant strategic retreat from Tesla’s earlier ambition to build a proprietary AI infrastructure. Peter Bannon, the lead of the Dojo team, is departing the company, and remaining members are being reassigned to other data center and computing initiatives within Tesla. The collapse of Dojo follows the exit of around 20 engineers who left to found DensityAI, a new AI startup set to emerge from stealth. The company is led by Ganesh Venkataramanan, former head of Dojo, along with ex-Tesla engineers Bill Chang and Ben Floering. DensityAI is developing AI chips, hardware, and software tailored for robotics, AI agents, and automotive applications—areas central to Tesla’s original Dojo ambitions. This shift comes amid growing challenges for Tesla’s self-driving vision. Despite Musk’s push to reposition Tesla as an AI and robotics company, the limited robotaxi rollout in Austin last June—where vehicles still required a human safety operator—was marred by erratic behavior and safety concerns. The project failed to deliver on the promised autonomy, undermining investor confidence. Dojo was conceived in 2019 as a custom-built AI training system powered by Tesla’s D1 chip, unveiled during the 2021 AI Day. The company aimed to use Dojo to process massive volumes of video data to train its full self-driving (FSD) system. A follow-up D2 chip was expected to address data flow limitations. However, development stalled, and by August 2024, Musk began promoting a new initiative: Cortex, a large-scale AI training cluster being built at Tesla’s Austin headquarters. Tesla is now pivoting toward external partnerships. The company is increasing reliance on Nvidia’s GPUs and AMD for compute power, while turning to Samsung for chip manufacturing. In a major move, Tesla recently signed a $16.5 billion agreement with Samsung to produce its AI6 inference chips—designed to power everything from FSD and Optimus robots to high-performance data center AI. At the second-quarter earnings call, Musk hinted at convergence between Dojo 3 and the AI6 chip, suggesting a unified architecture. “It seems like intuitively, we want to try to find convergence there, where it’s basically the same chip,” he said, signaling a shift from custom hardware to scalable, off-the-shelf solutions. The decision arrives as Tesla’s board offers Musk a $29 billion compensation package to keep him focused on the company’s AI and robotics ambitions, rather than diverting attention to his other ventures like xAI. The move underscores the pressure to deliver tangible progress in AI, especially as competitors advance rapidly in autonomous systems and generative AI. While Dojo’s demise signals a retreat from vertical integration, Tesla’s continued investment in AI—through partnerships and chip design—suggests the company remains committed to AI leadership, albeit through a more collaborative and pragmatic path.