Elon Musk has announced that Tesla will be reviving its work on Dojo3, a cutting-edge supercomputer project aimed at revolutionizing the company's Full Self-Driving software. The news comes as a surprise, given that the Dojo team was disbanded last year in favor of prioritizing AI chip development for onboard vehicles.
According to Musk, Tesla will restart work on Dojo3 once the AI5 chip design is finalized, citing its potential to process vast amounts of data from Tesla vehicles and enhance the company's neural net capabilities. The Dojo project initially focused on training the "neural net" behind Full Self-Driving software using video recordings and other data from the vehicles.
However, last year Musk stated that it didn't make sense for Tesla to divide its resources between two distinct AI chip designs, opting instead for a unified approach centered around the AI5, AI6, and subsequent chips. These chips are designed primarily for inference on board Tesla vehicles, not training. The AI6 chips will be manufactured by Samsung at Tesla's Texas factory under a $16 billion agreement.
Musk has also expressed interest in using space-based AI compute, proposing that data centers in orbit provide an alternative to land-based computing centers. Proponents of this idea believe that the advantages of space include easier access to solar energy and reduced power consumption due to lower temperatures. However, experts remain skeptical about the feasibility of such a project.
While the return to Dojo3 may signal Tesla's continued commitment to developing advanced AI technologies, it remains to be seen whether these efforts will yield the desired results or succumb to the same pitfalls that have plagued similar projects in the past.
According to Musk, Tesla will restart work on Dojo3 once the AI5 chip design is finalized, citing its potential to process vast amounts of data from Tesla vehicles and enhance the company's neural net capabilities. The Dojo project initially focused on training the "neural net" behind Full Self-Driving software using video recordings and other data from the vehicles.
However, last year Musk stated that it didn't make sense for Tesla to divide its resources between two distinct AI chip designs, opting instead for a unified approach centered around the AI5, AI6, and subsequent chips. These chips are designed primarily for inference on board Tesla vehicles, not training. The AI6 chips will be manufactured by Samsung at Tesla's Texas factory under a $16 billion agreement.
Musk has also expressed interest in using space-based AI compute, proposing that data centers in orbit provide an alternative to land-based computing centers. Proponents of this idea believe that the advantages of space include easier access to solar energy and reduced power consumption due to lower temperatures. However, experts remain skeptical about the feasibility of such a project.
While the return to Dojo3 may signal Tesla's continued commitment to developing advanced AI technologies, it remains to be seen whether these efforts will yield the desired results or succumb to the same pitfalls that have plagued similar projects in the past.
! Think about all the data those cars are racking up and how crazy it'll be to process it all with a supercomputer like Dojo3 
. I'm hyped that Elon's revisiting it after prioritizing AI chips – might just change the game for Full Self-Driving software 
. Plus, who knows what other tech innovations could come out of this? SpaceX's space-based AI compute idea is wild and would be so cool to see happen 
. Just got my fingers crossed that they can make it work without another major setback 
. Like, if Tesla was already all about the unified AI5 chip design, why revive Dojo3? It just feels like they're throwing money at a problem without really solving it 
. And what's up with the space-based AI compute thing? I get the idea of using solar energy and reducing power consumption, but isn't that just rehashing old tech? 
It sounds like Musk is trying to create this whole new level of innovation, but we all know how that usually goes down 
. I'm keeping a close eye on this one, though - if they can actually make it work, it could be huge for the future of AI and transportation. But so far, I'm not convinced 
.
like who wouldn't want to see a cutting-edge supercomputer go from being scrapped to being a game changer for full self driving? 
. after he banded the Dojo team last year, I thought it was a done deal. now suddenly he wants to revive it just because AI5 chip design is finalized? yeah right 
. what changed? was there some new data or something? 
also, have you seen the cost of those chips? $16 billion on Samsung's Texas factory? that's like, a whole lotta cash 
, we didn't know what a supercomputer was, but now they're talking about reviving Dojo3? I mean, it's like trying to fix a broken toy that you already threw away 
. But if it falls flat like all the other AI projects, I'll be saying "I told you so" 
.
! I was wondering what happened to Dojo3 and now it's back on the table 
. I'm all about supporting Tesla's innovative projects and the potential for space-based AI compute is MIND-BLOWING 
It might sound crazy, but Elon always pushes boundaries and I respect that 
. Let's keep our fingers crossed that this new project will be a success and take us to the next level in AI tech 
!
. I mean, last year they said they were ditching it for AI chip development, and now they're bringing it back? It seems like a bit of a U-turn to me 
. I'm still trying to wrap my head around this one...
But idk about the skeptics, maybe he's just gonna make it happen 
I've been thinking about this for a while now, and the idea of using solar energy to power computing centers in orbit just seems like the future. but like, we'll have to see how it all plays out...
.