NVIDIA (NVDA 3.06%) recently introduced Drive AutoPilot, a new reference platform that lets automakers add automated driving features to their vehicles, at CES 2019. The platform improves current advanced driver-assistance systems (ADAS) with features that bridge the gap between semi-autonomous vehicles and fully autonomous ones.
These features include lane changes, highway merging, pedestrian and cyclist detection, parking assistance, and personal mapping features. Inside the vehicle, it provides driver monitoring features and AI copilot capabilities. It also allows updates to its software over the air -- a feature most major automakers (except Tesla) -- haven't adopted yet. The platform is powered by NVIDIA's Xavier AI-focused SoC, which can process over 30 trillion operations per second.
Drive AutoPilot isn't a fully driverless platform like NVIDIA's Drive PX series of onboard computers (a version of which powers Tesla's similarly named Autopilot feature). Instead, Drive AutoPilot seems squarely aimed at Intel's (INTC 0.31%) Mobileye, which provides ADAS platforms for over 90% of the world's automakers.
Will NVIDIA's new platform significantly widen its moat against Intel, which is collaborating with BMW, Fiat Chrysler, and Delphi to launch fully autonomous vehicles by 2021?
Understanding NVIDIA's automotive ambitions
NVIDIA's Tegra series of ARM-based CPUs form the foundation of its driverless business. NVIDIA originally tried to market the Tegra as an application processor for smartphones, but it was pushed out of the market by Qualcomm. But instead of abandoning the chipset, NVIDIA pivoted the Tegra toward other markets -- including game consoles (like the Switch), drones, and navigation and infotainment systems for connected cars.
NVIDIA's early foothold in the connected car market led to the launch of Drive PX in 2015. The second version, Drive PX 2, was introduced in 2016. In 2017, NVIDIA introduced Drive PX Xavier and Drive PX Pegasus -- two platforms that were designed for fully autonomous vehicles. Last September, it launched Drive AGX Xavier, a scalable platform for autonomous machines, robots, healthcare devices, and self-driving cars.
NVIDIA currently has over 370 Drive partners, including Toyota, Mercedes-Benz, Audi, Volvo, and various service providers and auto suppliers. Those companies support the growth of its Automotive revenue, which accounted for 5% of its top line during the third quarter.
NVIDIA's Automotive revenue rose 19% annually and 7% sequentially to $172 million during that quarter. NVIDIA attributed that growth to a ramp up for next-generation AI-based cockpit infotainment systems as well as its ongoing efforts in the driverless market.
Unfortunately, the growth of NVIDIA's auto unit couldn't offset a sequential decline in its core gaming revenue, which tumbled as the cryptocurrency mining bubble popped and flooded the market with cheap GPUs. NVIDIA believes it will take one or two quarters for its gaming business to stabilize, so it's focusing on more promising markets like cars for the time being.
Gauging the challenges
NVIDIA's Drive AGX Xavier and Drive AutoPilot are clearly aimed at tethering more developers, companies, and automakers to its ecosystem. However, Intel remains a formidable rival in the automotive market for three reasons.
First, Mobileye already dominates the ADAS market, and it's leveraging that lead to sell its EyeQ computer vision chips for semi-autonomous and fully autonomous vehicles. Nearly 30 automakers already use its EyeQ chips, and the EyeQ 5 -- which is set to arrive in 2020 -- targets fully autonomous vehicles. This means locked-in automakers could stick with Mobileye's ADAS platforms instead of giving NVIDIA's Drive AutoPilot a chance.
Second, Intel can bundle Mobileye's ADAS platforms and EyeQ chips with its own Atom Automotive chips for infotainment and navigation systems. It could also mix in some of the technologies it gained from Movidius, the computer vision chipmaker it acquired in 2016.
Lastly, Intel's alliance with BMW, Fiat Chrysler, and Delphi is attracting a lot of attention. Unless NVIDIA matches Intel's might with a similar alliance (instead of its simpler Drive partnerships), it could struggle to reach more mainstream automakers.
The key takeaways
NVIDIA has a first-mover's advantage in the automotive space, but Intel is clearly catching up. The introduction of Drive AutoPilot indicates that NVIDIA recognizes the combined might of Intel and Mobileye -- and it likely hopes that its platform will win over automakers by leapfrogging over Mobileye's current-gen ADAS solutions.
This platform probably won't move the needle for NVIDIA's Automotive unit this year. However, Tier 1 suppliers Continental and ZF recently announced that they will launch semi-autonomous systems based on the Drive AutoPilot platform by 2020 -- so it's already sparking some early interest.