Electrek has reported that Tesla (NASDAQ:TSLA) will base its Tesla Vision replacement for Mobileye (NYSE:MBLY) on Nvidia (NASDAQ:NVDA) hardware. If this report turns out to be correct, it will be a huge win for Nvidia that will add to Nvidia's top line in the automotive segment, as well as probably lead to additional automotive customers. CEO Jen-Hsun Huang's positioning of Nvidia in machine learning and automotive driver assistance systems couldn't have been better timed.
The Benefits of Proximity
The Electrek article provides substantiation for my take on the Tesla/Mobileye breakup ("Did Nvidia Come Between Tesla And Mobileye?"). I had argued that if Tesla needed an alternative to the Mobileye hardware, then it couldn't do better than Nvidia's Drive PX 2.
Furthermore, there was evidence that Tesla had been working since the beginning of the year on a Mobileye alternative. In January, it was reported that Tesla had hired Jim Keller (a microprocessor engineer who had worked at AMD (NASDAQ:AMD) and Apple (NASDAQ:AAPL)) to be the head of the Autopilot Hardware Engineering. It was subsequently speculated that Tesla might develop its own system on chip (SOC), since this was Keller's specialty.
Keller probably waved off such an undertaking in the interests of time. With the Mobileye breakup imminent, he would have realized that Tesla needed a platform for which it could immediately begin developing the requisite image processing software.
According to Electrek, Tesla had already assumed responsibility for top level computer decision making for Autopilot, and only used Mobileye for image processing and object recognition. So it was not unthinkable that Tesla would seek to extend its responsibility end-to-end.
Furthermore, Nvidia already had a much more capable SOC in Nvidia's Parker, which had been specifically targeted at automotive autonomous vehicle and driver assistance applications. A pair of Parker Tegra SOCs are built into Drive PX 2, as well as a couple of Nvidia's most powerful Pascal GPUs. Together, they make Drive PX 2 the most powerful platform for driver assistance available today.
And yes, I've looked at NXP's (NASDAQ:NXPI) BlueBox Autonomous Driving Platform. After looking over the specs for the processors in the BlueBox, my professional opinion is that they probably don't come close to what Nvidia delivers in Drive PX 2. That's only an opinion, of course, based on the limited data available.
Another factor driving Keller's decision to go with Nvidia would be the geographic proximity of Tesla and Nvidia in the Bay Area. The importance of this should not be underestimated. I can personally vouch for the difficulty of managing an engineering effort by a company that's half way around the world. Keller would understand the benefits of collaborating with a company like Nvidia that was "just up the street."
Tesla's Driver Assistance "Leadership"
The Electrek article couldn't resist extolling Tesla's "leadership" in driver assistance. This undoubtedly will be challenged by Tesla bears who have tended towards a very negative view of Autopilot. However, Electrek was able to cite some reasonable objective testing performed by Motor Trend to back up its claim. In the performance tests MT performed, Autopilot was generally as good or better than systems provided by GM (NYSE:GM), Daimler (DDAIY), and Hyundai (OTCPK:HYMPY). And in one key area, automatic lane keeping, Tesla was clearly superior.
Of course, MT didn't test the scenario that led to the death of the Florida motorist, probably because all of the systems would have failed that test. But the tests do confirm that Tesla has a very competitive and well engineered offering in driver assistance, and is well ahead in the key area of lane keeping.
At issue is whether Tesla jeopardizes that leadership in making the transition to Nvidia from Mobileye. This is clearly a concern for Nvidia investors such as myself, since it has the potential to do more harm than good. Unfortunately, we just don't know enough about how Tesla is going to manage that transition.
Here are my expectations, based on my engineering judgment. Probably Tesla has been working on the Mobileye replacement based on Drive PX 2 since at least when Keller arrived in January. Probably, Tesla needs to have an Nvidia based replacement for Mobileye for inclusion in all its production cars by early 2017.
The first version of PX 2 to be included in Tesla cars would probably be based on the smaller, lower power version that Nvidia recently announced. Using just one Parker SOC, it would draw considerably less power than the full size version.
Is a year enough for the software development effort? It doesn't really feel like enough, but Tesla may simply have no choice given future unavailability of Mobileye hardware. The big uncertainty is when Tesla actually started the effort. It might have started it a year or two ago as a low level exploratory effort to see if a home grown solution could do a better job than Mobileye. If that's the case, then the effort stands a much better chance of success.
So, although there's considerable risk to Nvidia, there's also a big payoff. If Drive PX 2 starts to be installed in Tesla's production vehicles, Model S and X, and eventually Model 3, then Nvidia will have found a volume customer for Drive PX 2 for the first time. Instead of selling a few hundred units for development purposes, Nvidia would be selling tens of thousands every quarter.
To give you a feel for the revenue involved, here's a sample calculation. If Nvidia shipped Tesla 50,000 Drive PX 2 systems every quarter, that could add about $50 million in revenue to the Automotive segment. Last quarter, Nvidia's Auto segment had revenue of $119 million. If Tesla eventually reaches its goal of 500,000 vehicles per year, that's $125 million in additional revenue to the Auto segment every quarter.
And Nvidia will have established itself as the premier provider of driver assistance hardware and software. I doubt that Tesla would be Nvidia's only customer for long.
There are risks, of course, that the transition could be bungled. But although I've taken a negative view of Tesla's financial management, I've never doubted Tesla's engineering prowess. Or Nvidia's for that matter. The two companies, collaborating on the problem, represent a powerful combination.
I'm inclined to think that the collaboration will be successful. Both companies realize what's at stake, both realize that they have to get this right on the first try. A gap in availability of Autopilot in future production vehicles would be far preferable to a release of an inferior system.
Tesla could probably mitigate any Autopilot availability gap by including the Nvidia hardware during manufacturing, and providing an over-the-air software installation once the software development effort was completed. In any case, the final system has to be at least as good as Mobileye on day one.
I'm cautiously optimistic that with Nvidia's help, Tesla can pull this off. I remain long Nvidia and recommend it as a buy.
Disclosure: I am/we are long NVDA, AAPL.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.