Tesla's Dojo Supercomputer: Elon Musk's Ambitious Bid to Rival Nvidia in AI Computing

Elon Musk reveals photos of Dojo D1 Supercomputer cluster -- roughly equivalent to 8,000 Nvidia H100 GPUs for AI training

The Dojo D1 uses a system-on-wafer design to deliver impressive processing power for AI video training. Fresh off firing up the Memphis Supercluster, claimed to be "the most powerful AI cluster in the world," Elon Musk has now shared pictures of a supercomputer cluster that uses his own homegrown Dojo AI accelerators. He also announced on the Tesla earnings call that he would double down on Dojo development and deployment due to the high pricing of Nvidia's GPUs. Aside from the opening of the xAI facility in Tennessee, which aims to have 100,000 Nvidia H100 GPUs on a single fabric, Musk said that he will have Dojo D1 up and running by the end of the year. As Musk said, it would have the processing power of 8,000 of Nvidia's H100 chips, which is "Not massive, but not trivial either." Musk first unveiled the Dojo D1 chip in 2021 with a performance target of 322 TeraFLOPs of power. Then, in August last year, Tesla was spotted hiring a Senior Engineering Program Manager for Data Centers, which is usually one of the first steps any organization would take when planning its own data centers. Tesla also doubled its orders for the Dojo D1 the following month, which shows its confidence in its performance. By May 2024, it was reported that the Dojo processor was already in mass production. Now, it seems that the Dojo chips have already made their way to the States and into Elon's hands, and yesterday he shared pictures of the Dojo Supercomputer at their home in the data center. The Dojo chips are system-on-wafer processors with a 5-by-5 array. This means its 25 ultra-high-performance dies are interconnected using TSMC's integrated fan-out (InFO) technology) so they can act as a single processor and perform more efficiently than similar multi-processor machines. TSMC manufactures Dojo chips for Tesla, and Musk will run them alongside his Nvidia-powered Memphis Supercluster. However, while the Tennessee facility is owned by xAI and is primarily used for training Grok, the Dojo chips are more tuned for AI machine learning and video training, especially as they will be used to train Tesla's Full Self-Driving technology based on the video data gathered from Tesla cars. When Musk combined all the chips he has on hand, he said that he'd have 90,000 Nvidia H100 chips, 40,000 Nvidia AI4, and the Dojo D1 wafers running by the end of 2024. This substantial computing power shows how much effort and resources the billionaire is pouring into artificial intelligence.

Elon Musk doubles down on Tesla's Dojo supercomputer, aiming to compete with Nvidia

This story is available exclusively to Business Insider subscribers. Become an Insider and start reading now. Have an account? Log in. The chips quickly and efficiently perform difficult calculations to render images and graphics and are key to training and deploying AI. "I'm incredibly impressed by Nvidia's execution and the capability of their hardware," Musk said Tuesday. "What we are seeing is that the demand for Nvidia hardware is so high that it's often difficult to get the GPUs." Despite Musk's complaints that Nvidia chips are difficult to come by, the billionaire admitted in June to redirecting $500 million in the AI processors meant for Tesla to X earlier this year, saying the EV maker didn't have the space to turn them on. On Tuesday, Musk said Tesla must "put a lot more effort" into Dojo to ensure the company has the training capability it needs without having to rely exclusively on Nvidia. "We do see a path to being competitive with Nvidia with Dojo," Musk said. "We kind of have no choice." But competing with Nvidia could be easier said than done, given the company's massive head start in the field. While the tech company has only become a household name in the last year, Nvidia has been working on its signature chips for years. In an interview with NPR earlier this year, Stephen Witt, a writer who profiled Nvidia's CEO for The New Yorker, estimated the company has as much as a 10-year head start on any other chip maker in the industry. According to a June CNBC report, Mizuho Securities estimated that Nvidia controlled 70 to 95% of the market share of chips used for AI. After the Tuesday earnings call, in which Tesla reported missing its earnings targets, Musk posted photos of Dojo on X, saying it will have roughly "8k H100-equivalent of training online by the end of this year." "We've really got to make Dojo work," Musk said during the call. "And we will." Musk has been positioning Tesla as an AI company and has been hyping its Robotaxi; the Robotaxi event has been delayed until October of this year.

Elon Musk Forced To Compete With NVIDIA With AI Supercomputer Because Of High Costs, Tight Supply

This is not investment advice. The author has no position in any of the stocks mentioned. Wccftech.com has a disclosure and ethics policy. Tesla's is aiming to compete with NVIDIA in machine learning through its Dojo supercomputer, according to fresh remarks from Elon Musk. Musk shared the latest bit of details for his firm's AI plans at Tesla's earnings call earlier today after a financial report that missed Wall Street estimates. During the call, he bemoaned the tight supply capacity of NVIDIA's GPUs, which are widely viewed in the industry as the top performers for computing AI workloads. This has led Musk to conclude that Tesla has no choice but to compete with NVIDIA through its own supercomputer. Tesla's Dojo uses its in house D1 chip, and future variants will use an upgraded D2 chip, which the firm revealed in late 2022, well ahead of the start of the current AI wave. Over the past couple of years, Musk has shared a slew of new products for Tesla, including its manufacturing robot called Optimus and new vehicles. These have pushed its Dojo supercomputer in the background, and today's call saw a Tesla investor ask him about the supercomputer's future. AI is key to Tesla's future since it uses machine learning (a subset of AI technologies) to train its assisted driving platform called Full Self Driving (FSD). In response, while commending NVIDIA on its "execution and the capability of their hardware," Musk lamented that "the demand for NVIDIA hardware is so high that, uh, it's often difficult to get the GPUs." Musk is concerned "about actually being able to get state of the art NVIDIA GPUs when we want them," which necessitates Tesla to focus more resources on Dojo to "ensure that we've got the training capability that we need." This forces Tesla to "double down on Dojo" and places it on a path to "being competitive with NVIDIA, with Dojo," he added. Tesla has "no choice because the demand for NVIDIA is so high and the, the, it's, it's obviously their, their obligation essentially to raise the price of GPUs to whatever the market will bear, which is very high," he explained. Tesla shared the latest technical details for the Dojo supercomputer during its AI event in October 2022. Musk confirmed in June last year that Dojo was online and running useful tasks at Tesla data centers for months. An update shared by Tesla's AI division on X added that Dojo's hardware was production ready in July 2023, and it predicted that Dojo would reach 30 exaflops by February this year. One exaflop of computing equals roughly 300,000 NVIDIA A100 GPUs. Musk shared more details about Dojo on X earlier today when he outlined that Dojo 1 will have roughly "8k H100-equivalent of training online by end of year." These comments were accompanied by a graph in Tesla's investor deck for its Q2 earnings, which revealed that the firm aims to have AI training capacity roughly equivalent to 90,000 NVIDIA H100 GPUs by 2024 end. These H100s are accompanied by roughly 40,000 of Tesla's HW4 AI computers, added Musk.

Elon Musk doubles down on Tesla's Dojo supercomputer, aiming to compete with Nvidia | Business Insider India

Elon Musk is barrelling full steam ahead on Dojo -- Tesla's supercomputer project and attempt at competing with AI chip superstar Nvidia. The Tesla CEO and billionaire addressed Dojo during the company's disappointing second-quarter earnings call on Tuesday, doubling down on Tesla's commitment to developing the state-of-the-art system. Musk acknowledged he is "quite concerned" about Tesla's inability to access the top-tier Nvidia graphics processing units, or GPUs, because of skyrocketing demand for microprocessors within the tech sector. The chips quickly and efficiently perform difficult calculations to render images and graphics and are key to training and deploying AI. "I'm incredibly impressed by Nvidia's execution and the capability of their hardware," Musk said Tuesday. "What we are seeing is that the demand for Nvidia hardware is so high that it's often difficult to get the GPUs." Despite Musk's complaints that Nvidia chips are difficult to come by, the billionaire admitted in June to redirecting $500 million in the AI processors meant for Tesla to X earlier this year, saying the EV maker didn't have the space to turn them on. On Tuesday, Musk said Tesla must "put a lot more effort" into Dojo to ensure the company has the training capability it needs without having to rely exclusively on Nvidia. "We do see a path to being competitive with Nvidia with Dojo," Musk said. "We kind of have no choice." But competing with Nvidia could be easier said than done, given the company's massive head start in the field. While the tech company has only become a household name in the last year, Nvidia has been working on its signature chips for years. In an interview with NPR earlier this year, Stephen Witt, a writer who profiled Nvidia's CEO for The New Yorker, estimated the company has as much as a 10-year head start on any other chip maker in the industry. According to a June CNBC report, Mizuho Securities estimated that Nvidia controlled 70 to 95% of the market share of chips used for AI. After the Tuesday earnings call, in which Tesla reported missing its earnings targets, Musk posted photos of Dojo on X, saying it will have roughly "8k H100-equivalent of training online by the end of this year." "We've really got to make Dojo work," Musk said during the call. "And we will." Musk has been positioning Tesla as an AI company and has been hyping its Robotaxi; the Robotaxi event has been delayed until October of this year.