How can Tesla benefit from the AI boom? Do you remember Dojo?
Nvidia is one of the top AI hardware companies in the world. Even if Nvidia stops its R&D, a few years may not be enough for other players to catch up.
Jensen Huang, co-founder and chief executive officer of Nvidia Corp., speaks during the Hon Hai Tech Day in Taipei on October 18, 2023. (Photo by I-Hwa Cheng / AFP) (Photo by I-HWA CHENG/AFP via Getty Images)
From a recent Yahoo article:
Nvidia's rivals are circling, but they're still years from catching up“Software continues to be Nvidia's strategic moat,” explained Gartner VP analyst Chirag Dekate. “These ... turnkey experiences enable Nvidia to be at the forefront of mindshare, as well as adoption.”Nvidia’s lead didn’t happen overnight. It’s been working on AI products for years, even as investors questioned the move.“Nvidia, to its credit, started about 15 years ago working with universities to find novel things that you could do with GPUs, aside from gaming and visualization,” explained Moor Insights & Strategy CEO Patrick Moorhead.“What Nvidia does is they help create markets and that puts competitors in a very tough situation out there, because by the time they've caught up, Nvidia is on to the next new thing,” he added.
The next AI winner
Interestingly, the next AI winner may come from Tesla. Tesla’s Dojo system is a Tesla-designed supercomputer made to train the machine learning models behind the EV maker's self-driving systems. The computer takes in data captured by vehicles and processes it rapidly to improve the company's algorithms.
How does Tesla's Dojo system compare to Nvidia's A100 GPU? — Tesla's Dojo system has already surpassed the performance of Nvidia's A100 GPU and shows potential for even greater improvements, indicating a promising future for Tesla in the market.
Tesla’s Dojo supercomputer
Tesla Dojo AI Tile Microarchitecture
Inside Tesla's Innovative And Homegrown “Dojo” AI Supercomputer
This is extracted from a recent Yahoo Finance article about Dojo:
Tesla (TSLA) has been receiving a lot of attention over its new AI project: the Dojo supercomputer. The company has invested billions in advancing supercomputing architecture to efficiently and more accurately train its self-driving software.Dojo is designed to process vision and recognition data and input that information in Tesla’s machine learning models. The excitement surrounding Dojo stems from its ability to handle millions of terabytes of video data per second. Tesla projects that the Dojo will be in the top five most powerful supercomputers in the world by early 2024.The company’s goal is to effectively process this large amount of sensory data captured from real-life situations in over four million Tesla cars.
This is extracted from an article by Motley Fool.
Being a dominant force in AI requires having powerful supercomputers to train models, which is where Tesla's Dojo computer comes in. The first Dojo computer cost around $300 million to build and is powered by 10,000 GPUs. However, Tesla recently announced plans to build another one, expected to cost $500 million, at its Gigafactory in Buffalo, New York.
The Technology and scalability of Dojo
The following section is extracted from a Forbes article about the technological nuances and scalability of Dojo:
The architectural uniqueness of Dojo is evident in its building block, the D1 chip, manufactured by TSMC using 7 nm semiconductor nodes, with a large die size of 645 mm² and 50 billion transistors and leveraging a RISC-V approach and custom instructions. The system scales by deploying multiple ExaPODs, housing up to 1,062,000 cores and reaching 20 exaflops. This kind of scalability has never been more necessary, especially when one considers the gargantuan volume of data that Tesla's fleet generates. Furthermore, Dojo uses the software language PyTorch and introduces novel floating-point formats—CFloat8 and CFloat16—enabling more efficient vector processing and storage requirements.
From Indiana University post:
A 1 exaFLOPS (EFLOPS) computer system is capable of performing one quintillion (1018) floating-point operations per second. The rate 1 EFLOPS is equivalent to 1,000 PFLOPS. To match what a 1 EFLOPS computer system can do in just one second, you'd have to perform one calculation every second for 31,688,765,000 years.
Dojo boosts Tesla’s value
This processing power led to Morgan Stanley’s forecast of the total addressable market (TAM) of Dojo.
This is extracted from a September 2023 news article by Reuters of how Dojo can power the value of Tesla:
Tesla rallied 6% on Monday after Morgan Stanley said its Dojo supercomputer could power a near $600 billion surge in the electric-car maker's market value by helping speed up its foray into robotaxis and software services. Tesla, already the world's most valuable automaker, started production of the supercomputer to train artificial intelligence (AI) models for self-driving cars in July and plans to spend more than $1 billion on Dojo through next year. Jonas expects Dojo to drive the most value in software and services. Morgan Stanley raised its revenue estimate for Tesla's network services business to $335 billion in 2040 from $157 billion earlier. Jonas expects the unit to account for more than 60% of Tesla's core earnings by 2040, nearly doubling from 2030.
Conclusion
Tesla is a technology company with sustainable solutions in EV, solar, energy storage, and more. As a leading player in AI through its FSD, it has expanded its offering including insurance. While it is natural to look at Tesla via its EV sales, it is much more than an auto-maker. To beat Tesla, you can’t just have a better product, you will need a better factory.
With data being the new “digital gold”, Tesla could be one of the driving forces in computing prowess for the various data automation and AI applications.
Let us monitor and hopefully, we can get more updates.
Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.
