Electric vehicle leader Tesla Inc. (TSLA) has confirmed that it is developing its own specialized artificial intelligence software and hardware for its self-driving cars. Manufacturing the chips in-house should allow the Palo Alto, Calif.-based company to make them much more powerful at a fraction of the current cost, according to the company’s CEO, famed serial entrepreneur and investor Elon Musk. (See also: Tesla Sales to Beat $60B Within Decade: Nomura.)
On Thursday, at a company party following an AI conference in Long Beach, Calif., Musk broke the news that “Jim is developing specialized AI hardware that we think will be the best in the word,” referring to Jim Keller, Tesla’s vice president of hardware.
Keller noted that efficiencies gained by specializing and removing overheads between components could give Tesla’s chips 10 times the power at a 10th of the cost. Currently, the Silicon Valley automaker uses chips supplied by NVIDIA Corp. (NVDA). The announcement signals a move away from this partnership, though earlier this year the company was rumored to be working with NVIDIA rival Advanced Micro Devices Inc. (AMD).
Pluses and Minuses to AI
Musk also made a bold statement regarding the firm’s timeline for the deployment of self-driving cars, suggesting that fully autonomous vehicles will be on the road in the next two years. He gave the technology three years to reach “an order of magnitude better than humans.” Elon Musk, also the CEO and founder of SpaceX, Neuralink and the Boring Co., has recently called AI an “existential risk for human civilization,” and urging caution on the technology.
The news comes as the self-driving car space becomes increasingly crowded by the entrance of traditional automakers such as Ford Motor Co. (F) and General Motors Co. (GM), along with other tech giants including Alphabet Inc. (GOOG) and Apple Inc. (AAPL) and on-demand transportation players Uber Technologies and Lyft. (See also: Musk to Put His Personal Tesla Into Mars Orbit.)