Tesla, one of the companies of X (formerly Twitter) owning tycoon Elon Musk, is not only a manufacturer and seller of electronic cars.
The company wants to lead the race that many large technology companies are grappling with to impose their developments in Artificial Intelligence. According to analysts, Tesla has a significant lead over many of them.
All thanks to the development of Dojo, a supercomputer that the company has been building since 2021.. Its goal is to train, autonomously, Artificial Intelligence neural networks. In other words, to train machines to carry out more and more processes.
Dojo has moved from paper sketches to the production phase in July 2023. It is a supercomputer that, by using models of Machine Learningis capable of training Artificial Intelligence models applied to the driving of autonomous cars.
Becoming one of Elon Musk’s main personal bets – the executive has indicated that he will spend more than $1 billion next year for its creation – it is one of the most powerful computers in the world. By training thematic AI models, focused on a single niche, it is able to evolve much faster than a general application supercomputer.
This is because these computers are meant to process billions of data, huge amounts of information. In the case of Dojo, since all the information it processes is driving-related, it is able to generate models and patterns more quickly.
Practical application of Dojo in autonomous driving.
Dojo efficiently trains AI networks and machine learning algorithms for autonomous cars. Any AI, to operate efficiently, needs billions of data and very high processing power and capacity.
The more data it is provided with, the more processing power it needs. That’s why AIs need supercomputers that are capable of handling all that massive amount of data.
Tesla is using all the information it extracts from its supercomputer’s analysis of millions of videos and images to improve the self-driving function of Tesla vehicles. The more data it is provided with, the better trained the AI will be to be able to react to all kinds of situations.
Dojo will make it possible to improve the way an AI is trained. In this case, it applies to the self-driving of Tesla’s autonomous vehicles, but it could be applied to any other purpose. Dojo could use information to train other AIs in other sectors as well.
Dojo M1 Processor Chip
This is possible thanks to the proprietary chip that Tesla has developed, the Dojo M1 Chip, of superior data processing capability. Having its own processor – and not one from a competitor such as Nvidia – and such a powerful machine in its possession, gives Elon Musk – and Tesla – a privileged position in the race to create one of the world’s most important supercomputers. With equipment like this, with applications in any field, whoever is able to develop it and impose his model will gain a significant advantage over his competitors.
According to Elon Musk, Dojo could be fully operational by 2024. Its creation has lifted Tesla’s shares considerably – by up to 60% in recent weeks – and is expected to continue to do so over the coming months.
In fact, Morgan Stanley has considered that “Dojo” could meet all the requirements to be a real revolution and become the standard in the market. It processes visual data and makes it possible to solve autonomous operations, whether for vehicles, factories or, why not, even humanoids.