On Thursday, Bloomberg reported that Elon Musk, CEO of Tesla, has made the decision to dismantle the company’s Dojo supercomputer project. This development comes after the project’s leader and a significant number of its employees transitioned to a rival AI initiative, marking a considerable hurdle in Tesla’s ambitions to create a proprietary supercomputer for training AI models aimed at self-driving cars. The move raises concerns regarding Tesla’s efforts to keep its talent intact.
For years, Tesla has emphasized its custom supercomputer, designed to train machine-learning models for its Autopilot, Full Self-Driving, and Optimus robots, as a key advantage over its competitors. While many in the industry turn to external suppliers for computing power and chips, Tesla aimed to develop its own hardware, believing that a vertically integrated AI strategy would enable it to surpass competitors facing supply chain restrictions.
Technical delays and a talent exodus seemed to push success further and further away
However, persistent technical delays and the departure of key personnel have increasingly obstructed progress. Jim Keller, a renowned AI chip developer, left the company in 2018 shortly after being hired to spearhead Tesla’s chip manufacturing division. His replacement, Ganesh Venkataramanan, exited in 2023 to establish Density AI. Most recently, Peter Bannon, who took over the Dojo project after Venkataramanan, departed to join the same firm along with 20 other former Tesla engineers.
Other notable recent departures include Milan Kovac, head of engineering for Optimus, David Lau, who was VP of software engineering, and trusted Musk advisor, Omead Afshar.
Following the news from Bloomberg, Musk publicly acknowledged the shift in Tesla’s strategy, stating on X that the forthcoming AI chips for the company’s vehicles “will be excellent for inference and at least pretty good for training. All effort is focused on that.”
Reactions to this shift vary; some see it as an indication that Tesla’s AI ambitions have been exaggerated, while others view it as a prudent financial decision in a costly race to develop robotaxis and humanoid robots.
Regardless of the perspective, this decision marks a significant departure from Musk’s long-touted narrative concerning Tesla’s competitive edge.
Dojo was supposedly Tesla’s “secret sauce” for self-driving cars and humanoid robots
Gordon Johnson, founder and CEO of GLR Research, remarked that Dojo was expected to be Tesla’s “secret sauce” for enhancing the intelligence of its self-driving technology. Analysts once projected immense potential for Dojo, with Morgan Stanley even attributing a $500 billion valuation to Tesla asserting its future prospects on this initiative.
Moving ahead, Tesla appears poised to increase its reliance on external suppliers like Nvidia, AMD, and Samsung. The company recently initiated a $16.5 billion agreement with Samsung to secure chip supplies for upcoming EVs and robotaxis. Samsung is developing Tesla’s AI6 chips, with TSMC slated to provide the AI5 chips in the near future.
In the long run, investors might reward Tesla for stepping back from its Dojo project and reallocating resources, especially since competition in the EV market is intensifying and the removal of government tax credits weighs heavily on its core business. Notably, Tesla’s stock saw an increase of over 2.5 percent in early trading on Friday.
Nonetheless, the ongoing talent drain presents a significant challenge for Tesla, particularly with Musk seemingly competing against himself through xAI. The company faces difficulties that permeate almost every aspect of its operations. Musk’s controversial political views have led to challenges with the brand, causing many former supporters to distance themselves from the company. If further top executives depart, Musk may find himself steering Tesla’s AI ambitions alone.
