Tesla Inc.’s AI lead, Ashok Elluswamy, has spoken out about the company’s efforts to develop AI chips, a crucial aspect of its push into artificial intelligence and self-driving technology.
Chips Co-Designed With Tesla AI Software Teams
According to Elluswamy, the chips being developed by Tesla are “heavily co-designed with the Tesla AI software teams.” This co-design approach allows the chips to achieve better margins on performance per watt (perf per Watt) and performance per dollar (perf per dollar) than other AI hardware systems.
- Perf per Watt is a metric used to calculate a computer hardware’s energy efficiency, representing the amount of performance it delivers for the amount of power it consumes.
- Perf per dollar measures the computational power a system delivers for its cost.
Elluswamy highlighted that Tesla’s AI4 Chip can process and understand a million pixels of streaming video within approximately 1 millisecond. This feat is possible because Tesla’s software and hardware are designed together, allowing for optimal performance.
Elon Musk’s Bold AI Chip Prediction, FSD’s Final Puzzle Piece
Elluswamy’s comments follow Elon Musk’s claim that Tesla will produce more AI chipsets than any other chipmaker. Musk stated, “We expect to build chips at higher volumes ultimately than all other AI chips combined,” emphasizing the company’s commitment to its AI push.
Musk also hinted at the possibility of Tesla’s Full Self-Driving (FSD) technology achieving unsupervised autonomous driving. He mentioned that the upcoming FSD v14.3 update could be the final piece of the puzzle for this technology.
Tesla’s Robotaxi Push
Another significant development for Tesla’s autonomous driving efforts is the clearance of the self-certification process for its Robotaxi in the state of Nevada. This clearance could pave the way for commercial Robotaxi operations within the state.
Musk has announced plans to expand Robotaxis to Arizona, Nevada, and Florida by the end of the year. He also predicted that Tesla Robotaxis in Austin would go driverless by the end of 2025.
Key Takeaways
- Tesla is heavily invested in developing AI chips, with a focus on co-designing hardware and software for optimal performance.
- The company aims to produce more AI chipsets than any other chipmaker, emphasizing its commitment to AI technology.
- Tesla’s FSD technology is close to achieving unsupervised autonomous driving, with the upcoming FSD v14.3 update potentially being the final piece of the puzzle.
- The clearance of the self-certification process for Tesla’s Robotaxi in Nevada could lead to commercial operations in the state.