The gaming sector appears to be slower than investors had hoped for, and with Advanced Micro Devices (NASDAQ:AMD) becoming a stronger rival, the relative weakness of the gaming industry creates a key risk to Nvidia (NASDAQ:NVDA) stock. However, despite some of the negative commentary pertaining to Nvidia, there could be a silver lining to the recent GPU apocalypse. Perhaps, the on-going development within its ADAS (advanced driver assistance systems) division could offset some of the negative headwinds, as there were a number of wins recently tied to automotive OEMs, and on-going deployments of level 3-4 systems that could move the investor thesis forward in this area.
Rob Csonger (General Manager of Autonomous Machine Group Nvidia) was on the call with Timothy Arcuri (UBS Analyst), where they discuss the development of Nvidia’s ADAS systems, and also how they came up with their expected total addressable market opportunity:
Rob Csonger: The TAM that we talked about at our recent investor day – the $30 billion number — was based off of three components. One was a driving component; in other words this is revenue that NVidia would derive from computers and software that would go into the vehicles. The second component would be data center products, which are used for training and development of the software that goes into the car computers. And then finally at GTC we announced a new product called Drive Constellation, which is a simulation product for testing and validating.
Rob Csonger: So what we’re doing is – our technology of course is based on our newest processor called Xavier. And we are enabling the world’s automakers to deploy a system like this with even more performance and more functionality. And the announcement we made with Toyota is an example of that. So based on that, I think – you know, we estimated roughly 35 million cars and we also estimated that you would need two computers. You would need a computer for the selfdriving and you need a computer for the cockpit. And given a, you know — call it — hundreds of dollars, you know, ASP, you’d end up with, you know, a significant portion of that TAM.
What Rob Csonger is anticipating are a number of factors near-term that will drive the autonomous market forward. Namely, Toyota and Mercedes Benz, which have both expanded their collaboration on Nvidia’s autonomous platform, which means the shipment of a lot more Xavier chips when both companies are ready to launch their next-gen platforms (could be a couple years out).
Nvidia also released Xavier (several months ago) which does 30 trillion operations per second (heavily optimized for Nvidia’s custom CUDA architecture and machine learning algorithms). Rob Csonger also mentioned that they have “85 other companies developing on the Xavier platform.” Plus, it could have widespread adoption in robotics or other areas where datacenter on a chip makes sense. They started selling the developer kits (January 2019) for quantities of 1,000 (minimum) for $1,100 (implying that each developer kit will cost $1.1M per company at minimum).
Assuming, the automotive narrative plays out, Nvidia could reach a multiple billion/year revenue runway from the segment alone. Though, customers would have to value a more advanced autonomous driving system when buying their cars and opt for the upgrade. However, Tesla has a 77% attach rate for its advanced autonomous features, which cost $5,000 extra, according to Rob Csonger, so there may be meaningful adoption even if Toyota were to release the same feature.
If Toyota were to market a Level 4 system (which could work on freeways but probably not surface roads) and sold 1M cars with these features, while Nvidia generates $2,000 per unit (due to volume) they would generate an incremental $2B revenue from just Toyota alone. Toyota sold 10.4 million cars in FY’18, so they’d only need to sell the feature to just 10% of their customer base for it to become a viable volume business for Nvidia.
Autonomous remains Nvidia’s most promising growth area given the weakness in GPU shipments tied to gaming and crypto mining. The autonomous revenue might surprise in the next couple quarters as developer kit orders for Xavier starts to impact financial results. However, the adoption of Nvidia’s ADAS architecture might not translate into meaningful volume shipments for a couple years, but each auto OEM does intend to launch cars with ADAS level 4-5 systems in the next five-years.
Disclosure: The author has no position in Nvidia stock.
Read more on NVDA: