Self-driving cars are a major market driver right now, with multiple companies jockeying for a leadership position. Tesla is now working with AMD to build a specialized microprocessor to handle self-driving car technology. It’s also hired Jim Keller, the ex-AMD employee responsible for Ryzen’s design, as well as the K12 ARMv8 CPU that AMD never brought to market.
A self-driving vehicle win with a company as prominent as Tesla wouldn’t do much for AMD’s bottom line at this point, given Tesla’s low yearly volume, but it’s a potent PR win. The company’s share price spiked after the announcement, up 4.73 percent. The major loser in this deal is probably Nvidia, which has courted Tesla for years. In fact, it’s a little surprising to see Tesla pivoting to AMD after so many prominent moves with its biggest rival, and the implication Nvidia silicon like the Drive PX2 would be powering Tesla’s efforts for some years to come. Tesla’s design is apparently far enough along that the company has gotten silicon from GlobalFoundries and is running tests on it. That means the chip could be ready for integration within the next few months, assuming that the tests validate the design, of course.
Somebody’s having a good day.
There’s a few ways to read this. One is that whatever AMD is building for Tesla is entirely separate from the technology Nvidia provides, and that the company intends to use both firms. The other is that Tesla may be making a shift to AMD as opposed to Nvidia. Or, Nvidia might still provide chips for features like the infotainment console. Nvidia’s own website states:
Today, all Tesla vehicles—Model S, Model X, and the upcoming Model 3—will be equipped with an NVIDIA-powered on-board “supercomputer” that can provide full self-driving capability.
The computer delivers more than 40 times the processing power of the previous system, running an Tesla-developed neural net for vision, sonar, and radar processing.
This in-vehicle supercomputer is powered by the NVIDIA DRIVE™ PX 2 AI computing platform. It’s an end-to-end AI computing system that uses groundbreaking approaches in deep learning to perceive and understand the car’s surroundings.
“Supercomputer” marketing claptrap aside, Nvidia has been a prominent Tesla partner for at least six years. It’s unlikely to be thrilled to see AMD shouldering in on its turf, even if it still retains some of Tesla’s business. Tesla also recently disclosed that it was building a more powerful hardware model to bring self-driving vehicles to market. This would contradict Elon Musk’s promise that all vehicles since October 2016 have the hardware necessary to achieve “full self-driving capabilities,” but it might be a requirement if the car’s existing hardware can’t hack the problem. It’s extremely likely that whatever chip AMD built for Tesla is part of this effort.
Tesla, however, has downplayed the move. A spokesperson told The Verge, “This hardware set has some added computing and wiring redundancy, which very slightly improves reliability, but it does not have an additional Pascal GPU.”