Why do Wall Street's top investment banks believe Tesla Dojo is worth $500 billion?

Image source: Generated by Unbounded AI

On September 11, local time, Tesla's stock price surged 10.09% to close at US$273.58, with a market value of US$868.341 billion. Its market value increased by US$79.9 billion (approximately RMB 580 billion) overnight.

Tesla’s stock price surged due to a research report from Morgan Stanley. This report is optimistic about the market prospects of Tesla's Dojo supercomputer, and analysts believe it will increase Tesla's market value by up to US$500 billion (approximately RMB 3.6 trillion).

The report points out that the Tesla Dojo supercomputer can save Tesla up to $6.5 billion in costs and will accelerate the development of autonomous driving technology and humanoid robots, giving it a foothold in the autonomous driving market with a potential value of $10 trillion. Bring "asymmetric advantages".

Morgan Stanley said that Dojo allowed Tesla to move away from the controversy of "a car company or a technology company" and become more like a technology company.

So far, Musk has not commented on this matter, but has only forwarded a video about his new biography on X.

01 A research report, 80 billion US dollars overnight

On September 11, local time, a Morgan Stanley report predicted that Tesla's target share price would soar from US$250 to US$400 in the next 12 months, which means that the stock price is expected to rise by 60%.

This report's optimism about Tesla comes from the Tesla Dojo supercomputer. They believe that it has an "asymmetric advantage" in the autonomous driving market with a potential value of 10 trillion US dollars, which will bring Tesla up to 500 billion. The increase in market value of US dollars (approximately RMB 3.6 trillion) can save Tesla up to US$6.5 billion in costs and will accelerate the research and development of autonomous driving technology and humanoid robots.

Tesla stock price｜Source: Baidu screenshot

Morgan Stanley is very optimistic about the value of software services that Dojo can bring to Tesla, laying the foundation for Tesla to "take both software and hard" like Apple in the future. Morgan Stanley predicts that Tesla's network services business's revenue will increase to US$335 billion in 2040 from US$157 billion previously. Analyst Adam Jonas said in the report that Dojo can open up new potential markets and will "go far beyond selling cars at a fixed price." The unit will account for more than 60% of Tesla's core earnings in 2040 , doubled from 2030.

Lin Shi, an automobile analyst and secretary-general of the Intelligent Connected Vehicles of the China-Europe Association, said in an interview with a reporter from the Securities Times: "Tesla is actually committed to becoming an electric vehicle brand for civilians, and it has to achieve this through "scaling up." Its future layout does not rely solely on hardware to make money, but also on software, and on its APP and autonomous driving capabilities. This is its main profit direction in the future."

Dojo will also have an impact on Tesla's various businesses. Morgan Stanley expects Tesla's average revenue per user to increase from $100 to $180, an increase that takes into account additional revenue such as software services. In addition, shared travel business, third-party battery services and vehicle sales will also benefit from Dojo's improvement in autonomous driving and software services.

In addition, the recent "black swan" incident in the U.S. automotive industry may have a boost to Tesla's stock price. Currently, the three major U.S. auto giants - General Motors, Ford, Peugeot Citroën's parent company Stellantis and the United Auto Workers (UAW) are in a stalemate over the labor agreement. If the two sides fail to reach an agreement on a new labor contract before September 14, local time, a general strike of nearly 150,000 workers may be inevitable.

The strike has done huge harm to car companies. According to calculations by the Anderson Economic Group, a US consulting firm, if the strike against the Big Three lasts for 10 days, they will lose nearly US$1 billion. During the 40-day UAW strike in 2019, General Motors alone lost $3.6 billion.

Analysts believe that as traditional car companies are actively transitioning to electric vehicles, this possible strike may be a boon to Tesla. Wedbush analyst Dan Ives said in a research note that if a strike occurs, Tesla will have the opportunity to benefit from production shutdowns by competitors.

**02 Why can Dojo make Tesla money? **

Analyst Jonas pointed out in the report that investors have long debated whether Tesla is a car company or a technology company, and we believe Tesla is both. Dojo means Tesla is moving towards its goal of becoming a true technology company.

Morgan Stanley report｜Source: Internet

Dojo is a supercomputer developed by Tesla to train artificial intelligence machine learning and computer vision. Tesla uses videos and data from user vehicles to train this supercomputing system.

Tesla first unveiled Dojo at Tesla AI Day in August 2021. In July this year, Musk said at the company's financial report meeting that Dojo has begun production for training artificial intelligence models for self-driving cars. He is considering licensing its FSD hardware and software to other car manufacturers, and plans to invest more than 100 million yuan in Dojo by next year. $1 billion.

Since 2021, Tesla has been emphasizing the surprises brought by Dojo supercomputer in AI and autonomous driving. Dojo was originally designed to process large amounts of video data, accelerate the iteration of Tesla's Autopilot and Fully Self-Driving (FSD) systems, and provide computing power support for Tesla's humanoid robot Optimus. Dojo uses Tesla-designed chips and infrastructure and trains its neural network capabilities with video data from Tesla’s fleet.

Tesla already has a large supercomputer based on Nvidia GPUs, which is one of the most powerful supercomputer clusters in the world. The new self-developed Dojo supercomputer helps Tesla reduce its reliance on Nvidia GPUs.

Musk has said that the ultimate goal of the Dojo supercomputer is to achieve exaFLOP-level computing power per second, that is, it can perform tens of billions of floating-point operations per second, which will make it one of the fastest computers in the world. According to the Global Supercomputer Top500 list released this year, currently only the Frontier of the Oak Ridge National Laboratory in the United States has reached the exascale level, while the Japanese supercomputer Fugaku, which Tesla once set a target for exceeding, ranks second. .

Dojo may bring huge value to the autonomous driving business, which is its greatest potential. Jonas wrote in the research report, "If Dojo can help cars "see" and "react," what other markets can be opened up? You can imagine any terminal device with a camera that needs to make implementation decisions. He believes that self-driving cars have always been called the mother of all artificial intelligence projects. The advanced supercomputing architecture developed by Tesla breaks new boundaries of customized chips and may allow Tesla to compete in this potential $10 trillion industry. occupy a very high share of the market.

$10 trillion refers to the potential annual revenue from self-driving taxis when the vehicles are truly autonomous. Tesla currently plans to invest more than US$1 billion in the Dojo supercomputer project and complete the investment by the end of 2024. Musk has said that Tesla's market value is closely related to the perfection of its future self-driving technology. The potential of autonomous driving technology is that its value is very high. Even if only a small part of the technology is implemented, it will create huge returns.

The "asymmetric advantage" Dojo brings to Tesla in the autonomous driving market is that it will become Tesla's exclusive brain. To put it simply, self-driving cars need to collect an extremely large amount of data through sensors. The calculation and analysis of these data require ultra-high computing power, which is transmitted to the supercomputing system for continuous training and evolution. Among them, the Dojo D1 chip independently developed by Tesla has the ability to collect, train and evolve, and can more ideally collect information such as road traffic signs, biological images, road conditions, etc., and gradually realize FSD (fully automatic) through the analysis of these data. drive). These technologies can only be realized based on Tesla car terminals, thus creating brand barriers.

Tesla's humanoid robot Tesla Bot "Optimus Prime" also uses Dojo. Its head will be equipped with the same intelligent driving camera as Tesla cars, continuing the vision-based sensing technology line.

From this point of view, Dojo may become the computing infrastructure of Tesla's AI business, and may even target artificial general intelligence (AGI). Musk revealed in June this year that Dojo V1 is highly optimized for large-scale video training and is not designed for general AI purposes. However, Dojo V2 will break this limitation.

However, these are still looking at Tesla from the perspective of potential. At present, Tesla has not yet achieved autonomous driving, and the revenue from software services is not enough to have a substantial impact on Tesla's profit margins.