Nuked 
Hey followers! Today, we’re diving into Tesla’s surprising move to break up its Dojo supercomputer team and what it means for AI and self-driving tech.
Tesla has decided to disband its Dojo team, halting the development of its in-house AI training supercomputer. The project, which Musk described as crucial for full self-driving, was a mix of supercomputer hardware and chip design. Now, Tesla plans to lean more on external partners like Nvidia and Samsung for its AI chips.
Lead engineer Peter Bannon is leaving, and remaining team members will be reassigned within Tesla. This shift follows the departure of about 20 employees who started their own AI startup called DensityAI, focusing on hardware and software for AI data centers.
DensityAI, founded by former Tesla AI leaders, aims to build chips and systems for robotics, AI agents, and automotive use. This move comes at a critical time for Tesla’s AI ambitions, especially as Elon Musk emphasizes AI and robotics despite the company’s limited robotaxi rollout.
Musk has long touted Dojo as the backbone of Tesla’s AI efforts. However, recent developments suggest a strategic pivot. Musk hinted at focusing on Cortex, Tesla’s new AI training supercluster, which is under construction at Tesla’s Austin HQ. This signals a shift from the in-house Dojo project to external tech collaborations.
Tesla unveiled its D1 chip for Dojo back in 2021 and has announced a next-gen D2 chip. Now, the company is increasing reliance on Nvidia and AMD, as well as a recent $16.5 billion deal with Samsung for AI inference chips, indicating a significant change in sourcing hardware.
As Tesla’s board offers Musk a hefty $29 billion pay package, the focus remains on pushing AI efforts forward amid internal restructuring. The move to shut down Dojo reflects a strategic reassessment of Tesla’s long-term AI roadmap and chip development plans.
