Tesla's stock surged 10%! What happened?

Today, Tesla's stock suddenly surged 10%, and the reason behind it may represent a complete disruption of Tesla's business model. The excitement in the market is due to the latest report by Morgan Stanley on Tesla. The report upgraded Tesla from neutral to overweight and raised its target price from $250 to $400. So why did Morgan Stanley analyst Adams Jonas suddenly become so optimistic about Tesla? It’s all because of Tesla's supercomputer Dojo.

Dojo is a supercomputer developed by Tesla for training self-driving cars, which was unveiled for the first time in 2021. According to the company's statement, although Tesla already has many Nvidia chips as part of its training module, they are not enough, so the company decided to design its own chip that better meets Tesla's requirements for processing massive amounts of images every day. At last year's Tesla AI Day, the company announced progress on Dojo, including having a complete system tray and preparing to build modules. This summer, Dojo officially began production and is expected to reach 100 exaflops of computing power by the end of 2024, which can perform calculations at 10 to the power of 20 operations per second, potentially becoming the world's most powerful machine.

The reason behind this move is that Tesla's cars are equipped with sensors and make crucial decisions in extremely unpredictable driving environments. The ability of Tesla to improve the effectiveness of its FSD system relies on its ability to collect and process real-world video data at the edge, as well as training its robots using the experience of its large fleet of vehicles, which numbers around 5 million units today and is expected to reach closer to 50 million by the end of the decade.

To achieve this, Tesla requires a significant amount of compute power, specifically NVIDIA GPU clusters, but currently cannot acquire enough chips to properly train its vehicles. In addition, Tesla believes it can develop a more efficient system that caters to its specific needs while avoiding funding a supplier's 60% gross margin.

According to the report, analyst Jonas believes that using Dojo can greatly improve Tesla's AI training speed, making self-driving more promising, and also turning AI training from a cost into a profit point, becoming a SaaS model and leasing out its excess computing power. This will open up a $10 trillion market. Referring to Amazon AWS, they believe that Dojo will contribute 70% of Tesla's future profitability, which could bring the company $500 billion in value. On the cost side, Morgan Stanley's team expects that due to Dojo, the company could save $6.5 billion in costs over the next few years.

The chart above shows the breakdown of Morgan Stanley's $400 target price. It can be seen that the main impact of Dojo is on Mobility and Network Services, accounting for 86% of the target price increase. Mobility should mainly refer to Tesla's autonomous driving part, contributing $70 to the share price. While Network Services refers to the additional value generated by turning Dojo into a SaaS, contributing $59.

Jonas' argument is that Tesla's focus on its own AI chips will give the company a significant advantage. It is possible, but it could also be like Tesla's 4680 battery. In 2020, when Tesla showcased the 4680 battery, it felt revolutionary and would make Tesla far ahead of its competitors. But three years later, this never happened. On the contrary, China has made more battery innovations.