Morgan Stanley comments on Tesla's dissolution of the Dojo team: "DOGE-style efficiency" revolution begins, with a potential reallocation of billions in AI spending towards the robotics sector

According to Zhitong Finance APP, Morgan Stanley published a research report maintaining an "Overweight" rating on Tesla (TSLA.US) with a target price of $410. The report pointed out that recent reports indicate Tesla has disbanded its self-developed Dojo supercomputer team as an important measure to optimize the cost-effectiveness of its artificial intelligence (AI) program. The firm believes this strategic shift may deepen Tesla's collaborative efforts with Musk's xAI company, with related capital expenditures and R&D resources expected to be reallocated, focusing on breakthroughs in optimizing robot production costs and improving manufacturing processes.

Reports indicate that Tesla is disbanding its Dojo supercomputer team, with team leader Peter Bannon set to leave, although Tesla has not confirmed or commented on this news. The Dojo supercomputer was designed to process the vast amounts of data and video generated by Tesla vehicles for training full self-driving (FSD) and Optimus machine learning models. Sources revealed that Musk has ordered the termination of the project, and the company plans to increase collaboration with external technology partners such as NVIDIA (NVDA.US) and AMD (AMD.US) for support in computing, while also collaborating with Samsung Electronics (SSNLF.US) for chip manufacturing. Additionally, about 20 employees from the Dojo team recently left to join the newly established DensityAI, while the remaining members will be reassigned to other data centers and computing projects.

Morgan Stanley analyst Adam Jonas believes this move carries significant strategic and financial considerations. In his view, this may be part of Tesla's "DOGE-style" efficiency-driven cost-cutting plan aimed at reducing the costs of a project that could incur substantial capital and operational expenditures. Tesla's second-quarter report has shown that the AI program has increased operating expenses, with the training scale at the Texas Gigafactory expanding, and capital expenditures for fiscal year 2025 expected to exceed $9 billion—most of which will be directed towards AI-related fields.

The potential beneficiary of this strategic shift may be Musk's xAI company. Jonas noted that xAI is taking on an increasing amount of Tesla's "AI brain" development work while utilizing social media data from the X platform and real driving data from Tesla vehicles. The contraction of the Dojo project may pave the way for deeper collaboration between the two entities.

Jonas also observed that Tesla may be shifting its focus towards robotics technology and edge inference capabilities. Musk has repeatedly emphasized the "potential strategic value" of Tesla's global fleet as a distributed inference network. As the commercialization process of Optimus accelerates, analysts believe Tesla may redirect incremental capital expenditures and R&D investments towards reducing robot production costs and optimizing manufacturing systems—whether for autonomous taxis or humanoid robots.

The timing of Tesla's expenditure reduction is also intriguing. The GPU shortage dilemma that Musk has mentioned multiple times has significantly eased, eliminating a key constraint that previously forced Tesla to rush the development of self-developed computing resources like Dojo.

Furthermore, Morgan Stanley's U.S. semiconductor and NVIDIA analyst Joe Moore believes that due to NVIDIA's R&D investment exceeding $15 billion annually, and the expansion of AI investment into rack connectivity, software, and services, developing an ASIC chip that can surpass NVIDIA in mainstream tasks will become increasingly difficult At the same time, as AI tasks converge on complex reasoning, the cost-performance advantage of custom chips is disappearing