This article appears to be a collection of news snippets and updates about Tesla’s AI developments, particularly around the Dojo supercomputer project. Here are some key points:
Dojo Supercomputer
- Elon Musk mentions that Tesla is investing heavily in its own AI training capabilities through the Dojo supercomputer.
- The company aims to become competitive with Nvidia, which is a major supplier of GPU hardware for AI training.
Nvidia Supply Challenges
- Musk notes that demand for Nvidia GPUs is extremely high, making it difficult to secure supply.
- Tesla’s plan is to rely on its own Dojo supercomputer to ensure sufficient training capacity.
AI Training Capacity
- The company aims to increase its AI training capacity by roughly 90,000 H100-equivalent GPUs by the end of 2024 (up from around 40,000 in June).
- Musk also mentions that Dojo 1 will have "roughly 8k H100-equivalent" of training online by the end of this year.
Next-Gen AI Model
- Elon Musk reveals that current Tesla vehicles may not have sufficient hardware to support the company’s next-gen AI model.
- The new AI model has a roughly 5x increase in parameter count, which is challenging to achieve without upgrading vehicle inference computers.
