Elon Musk has shared additional estimates on the future of Tesla’s supercomputing cluster, Dojo, and highlighted roughly how much the company expects to spend on Nvidia products in 2024.
Discussion about Nvidia’s supercomputers arose on Tuesday after Tesla needed to have some of the company’s chips re-routed to X and xAI locations, due to a lack of storage, according to Musk. Following the reports, Musk later went on to post on X about Tesla’s purchases from Nvidia this year, which he estimates will be between $3 billion and $4 billion, along with sharing a few other insights about the company’s AI supercomputing needs.
Musk also noted earlier in the day that Tesla’s Giga Texas south extension is nearing completion, at which point it will house 50,000 Nvidia H100 chips used for FSD training.
Of the roughly $10B in AI-related expenditures I said Tesla would make this year, about half is internal, primarily the Tesla-designed AI inference computer and sensors present in all of our cars, plus Dojo.
For building the AI training superclusters, NVidia hardware is about…
— Elon Musk (@elonmusk) June 4, 2024
In a follow-up conversation to the post, Musk also responded to one X user asking if there was a “realistic pathway” toward Tesla producing its Dojo supercomputers at enough volume to rely more heavily on these for training compute than on Nvidia hardware. Musk responded by clarifying that training compute at Tesla is actually a small fraction of what’s needed, especially compared to inference compute needed for the FSD system.
“Training compute for Tesla is relatively small compared to inference compute, as the latter scales linearly with size of fleet,” Musk wrote. “Perhaps the best way to think about it is in terms of power consumption.”
“When the Tesla fleet reaches 100M vehicles, peak power consumption of AI hardware in cars will be ~100GW. Training power consumption is probably <5GW. These are very rough guesses.
“Obviously, 5GW of AI training compute is enormous by current standards, but is only about 5% of total Tesla AI compute.
While Musk calls it a “long shot,” he says it could be possible for production of Tesla’s Dojo to someday exceed that of Nvidia’s.
“There is a path for Dojo to exceed Nvidia,” Musk added. “It is a long shot, as I’ve said before, but success is one of the possible outcomes.”
Tesla self-driving development gets huge compliment from NVIDIA CEO
What are your thoughts? Let me know at zach@teslarati.com, find me on X at @zacharyvisconti, or send us tips at tips@teslarati.com.
Elon Musk
Tesla Full Self-Driving’s newest behavior is the perfect answer to aggressive cars
According to a recent video, it now appears the suite will automatically pull over if there is a tailgater on your bumper, the most ideal solution for when a driver is riding your bumper.
Tesla Full Self-Driving appears to have a new behavior that is the perfect answer to aggressive drivers.
According to a recent video, it now appears the suite will automatically pull over if there is a tailgater on your bumper, the most ideal solution for when a driver is riding your bumper.
With FSD’s constantly-changing Speed Profiles, it seems as if this solution could help eliminate the need to tinker with driving modes from the person in the driver’s seat. This tends to be one of my biggest complaints from FSD at times.
A video posted on X shows a Tesla on Full Self-Driving pulling over to the shoulder on windy, wet roads after another car seemed to be following it quite aggressively. The car looks to have automatically sensed that the vehicle behind it was in a bit of a hurry, so FSD determined that pulling over and letting it by was the best idea:
Tesla appears to be implementing some sort of feature that will now pull over if someone is tailgating you to let the car by
Really cool feature, definitely get a lot of this from those who think they drive race cars
— TESLARATI (@Teslarati) February 26, 2026
We can see from the clip that there was no human intervention to pull over to the side, as the driver’s hands are stationary and never interfere with the turn signal stalk.
This can be used to override some of the decisions FSD makes, and is a great way to get things back on track if the semi-autonomous functionality tries to do something that is either unneeded or not included in the routing on the in-car Nav.
FSD tends to move over for faster traffic on the interstate when there are multiple lanes. On two-lane highways, it will pass slower cars using the left lane. When faster traffic is behind a Tesla on FSD, the vehicle will move back over to the right lane, the correct behavior in a scenario like this.
Perhaps one of my biggest complaints at times with Full Self-Driving, especially from version to version, is how much tinkering Tesla does with Speed Profiles. One minute, they’re suitable for driving on local roads, the next, they’re either too fast or too slow.
When they are too slow, most of us just shift up into a faster setting, but at times, even that’s not enough, see below:
What has happened to Mad Max?
At one point it was going 32 in a 35. Traffic ahead had pulled away considerably https://t.co/bjKvaMVTNX pic.twitter.com/aaZSWmLu5v
— TESLARATI (@Teslarati) January 24, 2026
There are times when it feels like it would be suitable for the car to just pull over and let the vehicle that is traveling behind pass. This, at least up until this point, it appears, was something that required human intervention.
Now, it looks like Tesla is trying to get FSD to a point where it just knows that it should probably get out of the way.
Elon Musk
Tesla Megapack powers $1.1B AI data center project in Brazil
By integrating Tesla’s Megapack systems, the facility will function not only as a major power consumer but also as a grid-supporting asset.
Tesla’s Megapack battery systems will be deployed as part of a 400MW AI data center campus in Uberlândia, Brazil. The initiative is described as one of Latin America’s largest AI infrastructure projects.
The project is being led by RT-One, which confirmed that the facility will integrate Tesla Megapack battery energy storage systems (BESS) as part of a broader industrial alliance that includes Hitachi Energy, Siemens, ABB, HIMOINSA, and Schneider Electric. The project is backed by more than R$6 billion (approximately $1.1 billion) in private capital.
According to RT-One, the data center is designed to operate on 100% renewable energy while also reinforcing regional grid stability.
“Brazil generates abundant energy, particularly from renewable sources such as solar and wind. However, high renewable penetration can create grid stability challenges,” RT-One President Fernando Palamone noted in a post on LinkedIn. “Managing this imbalance is one of the country’s growing infrastructure priorities.”
By integrating Tesla’s Megapack systems, the facility will function not only as a major power consumer but also as a grid-supporting asset.
“The facility will be capable of absorbing excess electricity when supply is high and providing stabilization services when the grid requires additional support. This approach enhances resilience, improves reliability, and contributes to a more efficient use of renewable generation,” Palamone added.
The model mirrors approaches used in energy-intensive regions such as California and Texas, where large battery systems help manage fluctuations tied to renewable energy generation.
The RT-One President recently visited Tesla’s Megafactory in Lathrop, California, where Megapacks are produced, as part of establishing the partnership. He thanked the Tesla team, including Marcel Dall Pai, Nicholas Reale, and Sean Jones, for supporting the collaboration in his LinkedIn post.
Elon Musk
Starlink powers Europe’s first satellite-to-phone service with O2 partnership
The service initially supports text messaging along with apps such as WhatsApp, Facebook Messenger, Google Maps and weather tools.
Starlink is now powering Europe’s first commercial satellite-to-smartphone service, as Virgin Media O2 launches a space-based mobile data offering across the UK.
The new O2 Satellite service uses Starlink’s low-Earth orbit network to connect regular smartphones in areas without terrestrial coverage, expanding O2’s reach from 89% to 95% of Britain’s landmass.
Under the rollout, compatible Samsung devices automatically connect to Starlink satellites when users move beyond traditional mobile coverage, according to Reuters.
The service initially supports text messaging along with apps such as WhatsApp, Facebook Messenger, Google Maps and weather tools. O2 is pricing the add-on at £3 per month.
By leveraging Starlink’s satellite infrastructure, O2 can deliver connectivity in remote and rural regions without building additional ground towers. The move represents another step in Starlink’s push beyond fixed broadband and into direct-to-device mobile services.
Virgin Media O2 chief executive Lutz Schuler shared his thoughts about the Starlink partnership. “By launching O2 Satellite, we’ve become the first operator in Europe to launch a space-based mobile data service that, overnight, has brought new mobile coverage to an area around two-thirds the size of Wales for the first time,” he said.
Satellite-based mobile connectivity is gaining traction globally. In the U.S., T-Mobile has launched a similar satellite-to-cell offering. Meanwhile, Vodafone has conducted satellite video call tests through its partnership with AST SpaceMobile last year.
For Starlink, the O2 agreement highlights how its network is increasingly being integrated into national telecom systems, enabling standard smartphones to connect directly to satellites without specialized hardware.
Tesla Full Self-Driving’s newest behavior is the perfect answer to aggressive cars
Tesla Megapack powers $1.1B AI data center project in Brazil
