Tesla is building a new &#34;home&#34; for its Dojo supercomputer, which may one day help the company surpass other electric vehicle manufacturers.According to media reports on Tuesday, Tesla is constructing a new fortress-like building at its headquarters in Austin, Texas, which will house a part of the Dojo supercomputer. The supercomputer is being assembled by Tesla to assist in running the artificial intelligence (AI) software behind its autonomous driving features.Elon Musk has previously stated that eventually, Tesla may use Dojo to offer cloud services to other companies, similar to Amazon Web Services.This move reflects Musk's ambitious plan to have greater control over the technology required to operate the core AI software of Tesla's products. The Dojo supercomputer, based on Tesla's custom-designed D1 chip, could potentially enable Tesla to achieve autonomy on AI chips and reduce its reliance on Nvidia.Musk has expressed that Dojo has enormous potential and could bring returns worth hundreds of billions of dollars.In July of this year, Musk informed investors that Tesla plans to invest &#34;well over&#34; $1 billion in developing Dojo by the end of 2024. This supercomputer is the first machine Tesla started using at its data center in Palo Alto, California, this summer.Tesla currently uses Nvidia's AI chips to power its full self-driving software, which sends video data from the car's cameras back to the company for training the vehicle's systems to better recognize the surrounding environment.However, as Tesla expands its product lineup with new vehicles, the demand for AI computing power is expected to increase significantly.In April of this year, Musk told Tesla investors that Dojo has tremendous potential and could bring returns worth hundreds of billions of dollars:&#34;Dojo also has the potential to be a salable service, where we offer that service to other companies, similar to how Amazon Web Services started out just as a bookstore.&#34;&#34;If it turns out to be a good long-term bet, it's going to be a very, very big return. But the potential is there, yes, the potential is truly gigantic, like hundreds of billions of dollars.&#34;Dojo is also one of the reasons why some Wall Street analysts are bullish on Tesla's stock. Last month, Morgan Stanley analyst Adam Jonas predicted that Dojo could add $500 billion in enterprise value to Tesla. Jonas' prediction is based on the belief that Dojo can help Tesla achieve its ambitions in the robotaxi industry and bring in new sources of revenue beyond the automotive sector.According to reports, Jonas said that Dojo's customers may include companies using vision-based machine learning, including companies involved in robotics, aviation, railways, and facial recognition.One of the challenges for Tesla in these plans is that it is becoming increasingly difficult to obtain the chips needed to run supercomputers like Dojo, largely because the main supplier, NVIDIA, is unable to meet demand.Glenn O'Donnell, a semiconductor researcher at technology research firm Forrester, said:&#34;Everyone who tries to do this complains that their chips are in short supply.&#34;This is an important reason why Tesla is motivated to design its own AI chip, the D1. In fact, other major tech companies like Amazon and Microsoft have also started designing their own AI chips to reduce their reliance on NVIDIA.In July of this year, Musk stated that if chip manufacturer NVIDIA's supply can meet Tesla's ambitious goals, Tesla will continue to rely entirely on the company. He said that in the future, the company plans to continue using NVIDIA's chips while also using the D1 chip that powers Dojo.Last month, it was reported that Tesla plans to double the production of Dojo D1 chips next year, reaching 10,000 units, with TSMC as the foundry. Insiders revealed that it is expected that the order volume for TSMC's Dojo D1 chips will continue to increase until 2025.Cost savings are also another important factor in Tesla's decision to independently design chips.Jonas from Morgan Stanley wrote in the report that the D1 chip will allow Tesla to better control the power needed to run its AI software, and the company's adoption of the D1 chip for processing video data may be faster than using NVIDIA chips.Tesla has previously stated that it expects Dojo to enable the company to train fully autonomous workloads (discrete tasks or applications) in one week, a process that previously took a month, although these claims have not been confirmed. Morgan Stanley estimates that by designing its own chips instead of relying on NVIDIA, Tesla could save $6.5 billion in the next few years.Dojo can also run software for other AI products.To accommodate the development of Dojo, Tesla needs more and more physical space. The company is headquartered in Austin, with engineers also in California. The company has stated that it plans to use its Palo Alto data center to accommodate seven exapods, which is a Tesla term referring to a cluster of 10 refrigerator-sized cabinets filled with Dojo computers running D1 chips.Tesla predicts that in one year, its Dojo computing power will increase to the equivalent of 100 exapods. This growth may be one of the reasons why the company needs a &#34;second home&#34; for Dojo.According to the analysis, in the long run, Tesla's other businesses may drive additional demand for AI capacity.In July of this year, Musk told Tesla investors that the company had been in &#34;early discussions with a large OEM (Original Equipment Manufacturer) to license its Full Self-Driving (FSD) system.&#34; Musk did not specify which company it was, but it could be another commercial or consumer car manufacturer.He said:&#34;We don't want to keep this a secret.&#34;He was referring to Tesla's Full Self-Driving software.In his report on Dojo, Morgan Stanley's Jonas also suggested that Tesla could use Dojo to run the software behind its humanoid robot, Optimus, which is currently under development. He speculated that Musk's other companies, such as X and SpaceX, may purchase services from Dojo.

TSLA

Dojo D1 could help Tesla break free from its reliance on Nvidia chips. Musk stated that eventually Tesla may sell cloud services to other companies, similar to Amazon Web Services, using Dojo.

Dojo+D1, is Tesla planning to enter the AI cloud service market in the future?