This week NVIDIA (NASDAQ:NVDA) debuted a new single-processor configuration for its Drive PX 2 supercomputer. The company calls it a palm-sized, energy-efficient artificial intelligence computer for self-driving cars and mapping -- and it's another big step for the company toward its driverless car future.
Let's take a quick look at some of specifications of the processor, how the Chinese tech company Baidu (NASDAQ:BIDU) will use it, and why it all matters:
The new Drive PX 2 chip details
The new processor is a single-processor version of the company's Drive PX 2 supercomputer. NVIDIA offers its Drive PX 2 in multiple configurations, including single-processor, two processors with two discrete graphics processing units (GPUs), going all the way up to multiple Drive PX 2s in one car.
The new single-processor version is powered by the company's newest system-on-a-chip (SoC), with a GPU based on NVIDIA's new Pascal architecture. The processor also has an SoC specifically for handling inputs from multiple cameras, lidar, radar, and other self-driving component sources. (NASDAQOTH:BAMXF)
NVIDIA says the latest processor is specifically designed to provide highway automated driving for cars, as well as HD mapping, for the company's AutoCruise feature. And the chip packs some pretty powerful processing specs into a small configuration that consumes just 10 watts of power.
Baidu is already on board with the tech
NVIDIA has had versions of its Drive PX computers out for several years, but the latest iteration gives automakers and Tier 1 suppliers a more energy efficient version that still has strong artificial intelligence (AI) and neural networking capabilities.
And that's why Baidu will be using the new version in its HD map-based self-driving product for carmakers. The company has already done extensive driverless car testing in China, is even outpacing Google in some self-driving aspects, and plans on releasing autonomous vehicles in Chinese roads by 2018.
Baidu Vice President Liu Jun said that his company and NVIDIA are "leveraging our AI skills together to create a cloud-to-car system for self-driving." Baidu's choice of of the new Drive PX 2 chip means that NVIDIA will power the in-car computer in Baidu's cloud-based autonomous driving system.
What it all means for NVIDIA
NVIDIA's driverless car technologies are already used by 80 automakers and Tier 1 auto suppliers, and the new Drive PX 2 single-processor chip should help maintain that strong position.
NVIDIA said it will start shipping the new processor the fourth quarter of this year. The benefit here is that NVIDIA is consistently offering new configurations for automakers so that they can slowly introduce new driverless car features. Car makers are quickly bringing self-driving features into their vehicles, but it's still on an an incremental basis.
The new processor allows companies to easily add AI and deep learning processing to a vehicle, but without committing to too much technology all at once. NVIDIA should be able to benefit from this approach by offering automakers several entry points to the driverless car market.
NVIDIA already has a strong automotive business in selling GPUs to automakers, including Honda, Tesla, Volkswagen and BMW. Sales from that business segment equalled $119 million in fiscal Q2 2017. That represents just 8.3% of the company's total revenue right now, but don't think the auto market is a trivial pursuit for NVIDIA.
The driverless car market is expected to hit $77 billion by 2035, according to Boston Consulting Group. And IHS Automotive estimates that there will be 21 million cars sold each year with autonomous capabilities by 2035.
NVIDIA has positioned itself to tap into all that growth over the the next few years, and its early partnership with Baidu and other automakers will certainly help bring that about. NVIDIA investors may still have a little while to see the company's driverless car position pay off, but each new technology release brings that day just a little bit closer.