Elon Musk has always thrived on bold claims and the contradictions that follow. On 23 July, he joined Tesla’s second-quarter gain, the standard ritual in which executives talks about the performance and future of his company. Musk was in full vision mode.
That day, he painted a brilliant image of Dojo’s future, the ultra-ambitious, accustomed to Tesla supercomputer designed to train the neural networks for full self-drive (FSD) and Tesla’s humanoid robot, Optimus. He called the next generation of Dojo “really spectacular”, talked about a new “AI factory” with “a lot of potential”, and presented plans for Dojo 3.
His tone and detail suggested that Dojo was one of the most important Tesla projects, an active, prosperous pillar of the company’s identity as head of AI.
Less than a month later, Bloomberg Left bomb: citing sources familiar with the matter, the exit reported that Tesla had finished his project Dojo altogether. Peter Bannon, the head of the project, left the company, about 20 Dojo -Samideans recently left for a new start -up called Densityai, and the remaining staff was reassigned to other projects.
Complete around face
The speed of the reversal is difficult to exaggerate. In the course of weeks, Tesla seems to have gone publicly championing Dojo’s “spectacular” potential to quietly dismantle the team. If Bloomberg’s reporting is confirmed, it would mark the end of a Musk program once launched as central to Tesla’s technology edge.
The Dojo project has already endured turbulence. It has been widely seen as Tesla’s attempt to reduce dependence on NVIDIA GPUs building their own custom chips and training hardware. It was an expensive and technically risky strategy in a market governed by ragged giants. Losing talent to density, including key bosses, suggests mounting inner heads far in front of Bloomberg’s hand.
On the Julio CallMusk worked on a question of whether his starting AI, Xai, could use Dojo. His response was full of specifications and confidence.
“Dojo 2. We expect Dojo 2 to work in the following year. At about 100 kh, 100 counterparts,” Musk said.
He continued: “And then ai Five, who is also really spectacular. I do not use those words lightly. We hope to have the factory AI in production until the end of next year. But that has a lot of potential.”
Musk even looked forward, discussing “intuitive” convergence between Dojo 3 and chips used in Tesla and Optimus robots: “Thinking of Dojo 3 and the AI six in the first chip, it seems intuitively that we want you to have more than the or optimal. If you want high bandwidth wide communication between chips.
The field distortion of reality
Critics quickly captured Bloomberg’s report to attack Musk’s credibility, pointing out the contradiction between his recent Hype and the reported stop. The Tesla community in X (formerly Twitter) lit up with a debate. “As someone who is invested in TSLA over 10 years ago, I’m not good for his missed ambitious deadlines,” one user wrote. “But I think his comment on FSD, Optimus and Dojo for the past few years pass that. It’s misguided, but he knows it has to continue the collapse of Tesla’s car business.”
Another replied: “Like someone invested in TSLA for over 10 years, taking an article about Bloomberg about Tesla as all right of the first day is a bold choice. It could be true, but still bold. I suspect a specific part of Dojo has been canceled.”
As someone who has invested in TSLA for over 10 years, taking an article about Bloomberg on Tesla as quite right on day 1, is a bold choice.
It could be true, but still bold. I suspect that a particular part of Dojo has been canceled
– Michael Phippen (@Michaelphippen) 7 August 2025
For a long time Musk viewers, this is not new. In October 2015, Musk stated: “Tesla will have a car that can make full autonomy in about three years.” Three months later, he repeated: “We will end with complete autonomy,” adding: “And I think we will have complete autonomy in about two years.” Almost a decade later, Tesla cars are not yet fully autonomous.
The template – extreme optimism followed by sudden course changes – is a sign of what is often called a “reality distortion field”, a term first famous for Apple’s Steve Jobs. Musk used it to set aggressive deadlines for robotaxis, boring corporate tunnels and humanoid robots, often without delivering timely or scale.
If Dojo was actually torn only weeks after being presented as a near-future reality, the fall could be significant. Investors may question whether Musk consciously abused them in July, or whether a sudden, serious affair forced him to leave what he once described as a corner of Tesla’s AI strategy.
In all ways, if confirmed, the stop would treat another blow to Musk’s credibility, and deepen the question of where his vision ends and reality begins.
Tesla did not respond to a request for comment.
