Tesla is set to turn on its new NVIDIA-based AI-training cluster
NVIDIA
Today, Tesla is turning on its massive new system to advance its Full Self-Driving (FSD) technology. The system has many technicalities, so Tesla enthusiast and investor, Sawyer Merritt, shared insights on X.com.
Tesla uses Graphic Processing Units (GPUs) to train its neural networks that are used to solve AI-based problems. Training Tesla's AI involves using vast amounts of video to teach the system how to operate a vehicle autonomously and safely. NVIDIA, a leading GPU manufacturer, recently released a new chip, the H100, which offers significantly higher performance than its predecessor, the A100. However, the H100 doesn't come cheap. NVIDIA's latest training-focused GPU costs about $30,000, and Tesla needs more than just a handful.
Tesla's new training cluster will be powered by 10,000 H100 NVIDIA GPUs, bringing its cost to hundreds of millions of dollars for the GPUs alone.
However, due to the high demand for this advanced chip, NVIDIA cannot produce enough H100 units to meet Tesla's, and the industry's growing demand.
The Power of Dojo
As a result, Tesla is investing over $1 billion to develop its own supercomputer, Dojo. Dojo will have a custom-designed chip to train Tesla's neural nets that will be leveraged in FSD and Optimus. Elon Musk, Tesla’s CEO, indicated that if NVIDIA could supply enough chips, the creation of Dojo might not have been necessary. Yet, the current situation has made it imperative for Tesla to take matters into its own hands.
Merritt underscored that the computational power required to train Tesla's FSD technology is a significant challenge. However, Tesla is actively addressing this issue. Musk announced that the company plans to invest over $2 billion in 2023 and again in 2024 to enhance its computational capabilities. This investment is necessary to develop FSD technology and signals Tesla’s commitment to innovation and leadership in the automotive industry.
Groundbreaking Data Centers
Tesla is not only developing its own supercomputer but is also in the process of designing what it describes as its "1st of its kind Data Centers." A recent job listing for a Senior Engineering Program Manager in Austin, Texas, where Tesla's Giga Factory operates, reveals the complexity of this initiative.
The establishment of these data centers is integral to Tesla's broader vision. The company’s need for extensive computing resources is paramount, especially when processing and analyzing large amounts of video data for its self-driving software. The Dojo supercomputer is already a testament to their commitment in this area.
Subscribe
Subscribe to our newsletter to stay up to date on the latest Tesla news, upcoming features and software updates.
In this article, we’ll cover Tesla’s updates on Optimus, batteries, and Tesla Energy.
Optimus
Tesla has been working away on their humanoid robot and continues to make progress in software and hardware.
First, Tesla is preparing the Fremont factory for the Optimus pilot production line, which is scheduled for completion later this year. Once it is, wider deployments of Optimus for internal use within Tesla’s facilities are expected as well. Tesla aims to have several thousand Optimus units working in its North American factories by the end of the year once the pilot production line is operational.
Tesla’s goals for production remain extremely lofty - 1 million units per year by 2030. However, they could face some challenges when ramping production.
Key components like the shoulder actuators use specialized permanent and rare-earth magnets, which are currently sourced from China. Due to recent Chinese restrictions on the overseas sale of these magnets, Tesla is seeking an exemption or alternative suppliers. They have not yet looked into modifying the shoulder actuator but will likely do so if they cannot obtain the necessary materials.
Batteries
Batteries are another item that Tesla’s teams have been working on behind the scenes for years now. The second generation of the 4680 - the Cybercell - has been IRA-compliant for some time now. This means that the Cybertruck is eligible for the US Federal EV rebate.
Tesla also achieved the lowest cost-per-kWh of any of its cells with the 4680 battery - and it is potentially one of the cheapest cells being manufactured by any vehicle battery manufacturer at this point. With dry-cathode still being worked on, Tesla may be able to squeeze more optimizations and cost efficiencies from the 4680 cells.
Additionally, Tesla is progressing with its plans for lithium refining and cathode production in the US, both of which are scheduled to commence in 2025. While the company says they’re no longer supply-constrained for non-LFP vehicle batteries, on-shoring production and sourcing critical minerals from nations outside of China will be key.
LFP batteries continue to be supply-constrained, namely for the Tesla Energy division. LFP batteries and their materials are sourced from China. Due to tariffs and limited exports, Tesla can’t obtain enough and is considering potentially building an LFP production facility in North America.
Energy
Tesla’s energy division is still experiencing some of the highest growth of any of its divisions. Year over year, Tesla saw a 154% increase in energy storage deployments, including both Megapack and Powerwall - for a total of 10.4 GWh deployed in just Q1 2025. While deliveries in energy storage remain volatile due to the nature of Megapack installations, Tesla expects growth to continue rapidly in this segment.
Tesla also deployed 1GWh of Powerwall 3 residential storage this quarter, marking its strongest quarter. Powerwall 3 has received positive feedback from customers, many of whom appreciate its new capabilities with its built-in inverter for solar.
Megapack is continuing to see demand increases, currently highlighted by utility-scale Megapack systems, as well as data centers requiring stable power delivery. Megafactory Shanghai is also online now and producing Megapacks - with an annual production capacity of 20GWh today and up to 40GWh in the future. The site has also produced over 100 Megapacks this quarter, which are all awaiting delivery.
There was a lot of interesting news from Tesla’s Q1 2025 Earnings Call, covering everything from FSD and Robotaxi - to the less glamorous but equally important Megapack and Powerwall.
Tesla is heavily leaning into artificial intelligence, and its insurance offering is just another example of how it’s improving its product or lowering costs by leveraging AI.
Tesla recently started offering an insurance discount in select states when drivers use FSD for at least 50% of their drives and now it’s introducing an AI to help handle customer claims.
Tesla has developed an in-house voiced AI agent that can assist customers in handling simple support requests for Tesla Insurance.
For customers calling in from those states, the new AI agent provides a unique way to address the most common support calls. And it’s not just answering common questions but actually making requested changes to the owner’s account.
Policy Changes
The first key item is that it automates policy changes. Simple policy updates, including adjusting your deductible or coverage limits, are now done via AI. For policyholders who are simply looking to make quick changes and don’t have any questions, this makes the process a lot quicker by not having to wait for a representative. Tesla isn’t eliminating representatives, but this could reduce the number of representatives required or reduce wait times.
Continue Where You Left Off
The second item here, highlighted by Raj Jegannathan from Tesla’s internal IT team, is that Tesla’s AI agent is able to offer summaries of the user’s last interaction with Tesla Insurance. It will summarize your last interaction and provide assistance on that particular topic if you need to continue it. That means that you don’t have to wait for a human to review your file - the AI will kick off right where you left off.
Tesla appears to be focused on improving efficiency and making support more accessible. While actual items like claims are left up to humans due to their inherently complex nature, this helps free up employees to handle more complex items. While there’s no doubt Tesla will continue to develop this AI like they do everything else, we may soon see it take on even more tasks.
More AI
This isn’t the first AI agent that Tesla has demoed - there is now a chat-based AI sales agent available on the front page of Tesla’s website, which is able to answer common questions on Tesla vehicles.
Tesla has also been improving their AI support tool available in the Tesla App is able to provide feedback on common issues and also guide users towards either solving the problem or placing a support request.
Tesla’s strategy here is to influence the cost-heavy areas associated with having humans address simple requests and instead leverage AI, which can offer instant answers and reduce support costs.
Roll Out to More States
While this new AI is currently limited to just 12 states, it is likely to follow Tesla Insurance’s expansion. Insurance seems to have been at a bit of a standstill lately. Tesla continues to improve features such as the improvements to Safety Score V2.2, but we haven’t seen Tesla roll out support to new states since it added Minnesota in November of 2022.