Why Does Tesla Want to Make Its Own Silicon Chips?

In general, Tesla’s artificial intelligence We see that there are significant advantages in terms of The fact that Tesla pointed out that it will produce its own chips at the Tesla AI event held recently, shows that this advantage is more than expected.

It was Tesla’s humanoid robot and dance that gathered all the spotlights at the AI ​​event. Although it is overshadowed by the robot, one of the important developments belongs to the company itself. D1 It was an artificial intelligence chip. What could be Tesla’s purpose in producing chips?

reach the optimum

When comparing the features of Apple devices with the features of their competitors, the features of competing products are usually better on paper: more RAM, faster processor, etc. is in the form. A similar situation exists between game consoles and computers. However, both Apple products and consoles are much better at what they do than their competitors with equivalent features. efficient and powerful It is possible. Because they do not serve a wide area like the products used by their competitors, they are designed to be used in the most optimal way.

Head of Tesla’s artificial intelligence division Andrej Karpathy, in a previous statement, compared the autonomous car to a synthetic animal and said that the vehicle would develop accordingly. Elon Musk also took a few days to learn artificial intelligence thanks to the new chip in machine learning at the AI ​​event. it will be down in a few hours And he said it was important. So Tesla is trying to find a similar optimum point.

Everyone’s after chipping

On the other hand, only those who roll up their sleeves to make chips Tesla not. Although Intel is currently the market leader, Google, Amazon, Microsoft and other companies are also working on producing their own special chips. What makes this market attractive is that technology has entered almost every field of production and products.

Tesla D1 chips Tesla’s Dojo supercomputer will be part of the system. Chips with a 7-nanometer architecture will also be used for the Dojo. 25 of these chips will be placed in a single training frame. 120 squares come together to form a server.

Tesla’s aim is rather expensive technologies like lidar. a more affordable alternative to create. For this, it focuses on artificial intelligence with a neural network. Their aim is to establish systems that will process data from radars and sensors and train vehicles, thanks to optimally functioning chips.