Tesla Disbands Dojo Supercomputer Team

https://www.bloomberg.com/news/articles/2025-08-07/tesla-disbands-dojo-supercomputer-team-in-blow-to-ai-effort

by mfiguiere • 12 months ago

51 47 12 months ago

bloomberg.com

https://archive.ph/Rv4Bj

toomuchtodo• 12 months ago

https://archive.today/i1EAP

TheAlchemist6• 12 months ago

Just for fun, here is a great read from a "leading Tesla analyst" 2 years ago "Tesla jumps as analyst predicts $600 billion value boost from Dojo":

https://www.reuters.com/business/autos-transportation/tesla-...

Also worth noting that Elon said 2 weeks (!!!) ago on investors call that they expect to have Dojo operating sometime next year at scale equivalent to 100k h100... How people still don't see that this guy is just lying all the time is beyond me.

The ship is sinking very fast folks.

ethagknight2• 12 months ago

Given that AI is core to Tesla's robotic and self driving future, it seems like theres a lot more to this story than what has been leaked thus far.

iszomer• 12 months ago

Maybe an intentional troll, given the profile's project.the.snowball moniker. /shrug

cosmicgadget2• 12 months ago

There are other hardware vendors.

TheAlchemist6• 12 months ago

Yeah, there is more to the story - it's just pure fantasy !

Tesla robotic future ? They are years behind competitors... they are very good at faking autonomous robots though - people are still falling for their Twitter posts, Robotaxi unveil with teleoperated robots and Tesla Diner teleoperated robot serving popcorn (which last something like 2 days... then it was gone already). It's all PR stuff, nothing is really happening behind the scenes.

ivape• 12 months ago

Two years ago Elon actually sounded a warning about Dojo. I recall him mentioning this in an earnings call:

“And we’re pursuing the dual path of Nvidia and Dojo. But I would think of Dojo as a long shot. It’s a long shot worth taking because the payoff is potentially very high. But it’s not something that is a high probability. It’s not like a sure thing at all. It’s a high‑risk, high‑payoff program. Dojo is working, and it is doing training jobs, and we are scaling it up, and we have plans for Dojo 1.5, Dojo 2, Dojo 3, and whatnot. So I think it’s got potential, but I can’t emphasize enough: high risk, high payoff.”

https://thetranscript.net/transcript/5550/tesla-q4-2023-earn...

In other words, he admitted it’s a long shot to replace Nvidia with their own chip awhile ago. It’s interesting because Google hasn’t given up on their TPUs, but Elon being a hardware guy mostly, has.

kranke155• 12 months ago

I doubt Elon had the time or the inclination to build yet another hardware stack, and he probably saw something that showed him he couldn’t quite catch up to Nvidia?

The scale of Nvidia manufacturing and RnD should be staggering atm. Google built the entire software and hardware stack for TPUs before the AI boom. I can’t even imagine how hard it would be to hire GPU hardware talent right now.

mrandish• 12 months ago

it’s a long shot to replace Nvidia with their own chip awhile ago

It's also worth mentioning that having their own ambitious effort in secret development was a valuable BATNA (negotiating tactic) for long-term contracts with outside suppliers. The value of that recent >$16B deal with Samsung for AI chips thru 2033 may have benefited from the prospect of Dojo as a potential in-house alternative.

bamboozled• 12 months ago

It’s amazing how people are operating with such immunity. If I lied to my boss like that I suppose I’d just be fired and let go without pay…or would I get promoted ?

I’m not into fucking society over like these sociopaths, I guess that’s “my problem”.

enslavedrobot• 12 months ago

This is what people said in 2018 when the stock was 16$. In case you aren't keeping up Tesla signed a $17billion agreement with Samsung to produce their AI6 chip family which is designed to replace the dojo architecture. It is a harmonized design that can be scaled to training clusters or deployed in edge inference.

moralestapia• 12 months ago

Another Karpathy initiative that turned out to be utter trash ...

https://www.hpcwire.com/2021/06/22/ahead-of-dojo-tesla-revea...

DoesntMatter2210• 12 months ago

I don't understand co.ments like this. Clearly they realized that it wasn't worth doing and cut costs accordingly. This seems smart. Companies constantly place bets and adapt when they aren't working out.

As far as the ship sinking that's pretty silly. They continue to be profitable with some sectors like energy growing very quickly.

Virtually all EV makers have faced a couple of tough years. Many like Rivian and lucid are getting clobbered.

Tesla has a 37 billion dollar bank account and low debt. There is no scenario that the ship is sinking

TheAlchemist6• 12 months ago

What I'm pointing to is the absurdity of the situation. It was supposed to be a huge project, valued by some analysts at $600 billions, and years in the making. The CEO himself self recently they are on track to bring it online next year at a huge scale. 2 weeks ago he was still saying it, now it's axed.

It was clearly nowhere near where they said it was, or they would have some significant sunk costs and write downs. None of which exist in their financial statements for Dojo.

As for the ship sinking - I know it seems like a huge exaggeration now. But it's a car maker, with a collapsing sales and which made virtually all the profits from direct and indirect subsides - which just got axed too. That alone would be very worrying. But it's also a company which sold billions of $ of software that is still not working as advertised and lawsuit are only starting to come in. By their own admission the software won't work on >50% of cars they already sold and they would need to upgrade them 'sometime in the future'. Oh and the CEO is the most hated man in the US currently.

AtlasBarfed3• 11 months ago

"they continue to be profitable"

The credits are drying up, the sales crashed, the brand is dead, and there's only so many mark-to-market accounting you can do on cryptocurrency.

If the stock wasn't buoyed by some bizarre modern financial techno cult, I'd short it

techpineapple• 12 months ago

I wonder if this is to shift the gpus to X.AI?

KiwiJohnno• 12 months ago

I was thinking maybe this, but more likely X.AI would develop the models for Tesla, such as for FSD and Optimus.

Honestly, it might not be a bad move. X.AI has done some great stuff, Grok is rough around the edges but pretty much at the bleeding edge and it doesn't make a lot of sense for musk to be running two AI camps (Yes I know they are theoretically separate companies...(

jeffbee2• 12 months ago

I want to write a browser extension that analyzes past HN archives and decorates the usernames of posts and comments with a "Fell For It Again" badge in the case of people who seemed to believe that Dojo existed.

snypher• 12 months ago

Maybe you should provide a list of things you were wrong about so we can tag you accordingly.

jeffbee2• 12 months ago

Look, if you're talking about making the HN-sentiment-badge extension general and customizable, I am in.

Zigurd4• 12 months ago

What a h8er! Next, you're gonna tell me there's no actual Cybercab and those were left over prototypes of a failed attempt at a low cost vehicle, as if this was some kind of pattern of behavior,

cosmicgadget2• 12 months ago

Tesla has faced an exodus of key talent this year as it has grappled with rising competition, falling sales and a consumer backlash to Musk’s political activity. Milan Kovac, the head of engineering for Optimus, and David Lau, vice president of software engineering, departed earlier this year, while Bloomberg reported in June that longtime Musk confidant Omead Afshar had abruptly left.

Wonder how much is belt-tightening from poor financials and how much is talent that was tempted/driven away.

Zigurd4• 12 months ago

Elon is very much like that boss who starts too many projects, under-funds all of them, and then gets pissy about under-resourced projects failing to produce exceptional results. Just three or four orders of magnitude scaled up from the typical tech industry leader wanna be who does this.

DoesntMatter2210• 12 months ago

Let's see here. Tesla was the first profitable electric car company. It is still profitable.

Space X is also profitable and growing quickly. Boring company is doing surprisingly well having done the loop in Las Vegas and now doing a big one in Tenn.

Neuralink has 9 working patients at this point who have been helped tremendously.

xAi has the premier model for deep though.

So "failing to produce exceptional results" doesn't seem to add up.

They are smart about shutting down things that won't work. That is smart

AtlasBarfed3• 11 months ago

Elon motivated people to do things, it was one of steve jobs superpowers.

The naxi salutes had to have killed that. Yeah he'll find zealots and mouthbreathers, but talented ones?

Those companies did it with a passion. Not from the CEO, but the employees.

paxys• 12 months ago

"Tesla builds Dojo Supercomputer" - stock goes up

"Tesla disbands Dojo Supercomputer" - stock goes up

thejazzman• 12 months ago

now that just sums it all up beautifully. over and over again.

rbanffy• 12 months ago

The packaging they were planning to use was very interesting: https://www.tomshardware.com/tech-industry/tsmc-to-go-3d-wit...

Tesla Disbands Dojo Supercomputer Team

Related Stories