A lot of the leading car manufacturers in the world are playing catch up with Tesla. I believe this is because many of them simply didn’t believe that Tesla could deliver. But if you look at Tesla’s volume for the Tesla Model 3, there are some significant proof points that the company has. In fact, in December, the Tesla Model 3 was the third best-selling car in all of Europe. Tesla is not just eating into European car manufacturers’ hybrid sales, but also into their gas-powered vehicles as well. The automotive industry isn’t just made up of the auto manufacturers, it also includes suppliers and their partners as well, and in Germany this ecosystem is arguably the biggest.
How did we get here?
Many of the car manufacturers simply wrote Tesla off as a company that couldn’t execute and one that would never be successful. That was especially true of the Model 3, which has long been the company’s strategy for reaching a broader market and even more important, profitability. Tesla struggled mightily to deliver the Model 3 with numerous setbacks and stumbles, but eventually the company figured things out and started shipping Model 3s in volume and has shipped 400,000 vehicles thus far. Nobody really expected that Tesla would be able to ship 400,000 Model 3s by 2020, especially the European auto manufacturers. So much so, that both Mercedes and BMW have had significant shakeups including layoffs in the case of Mercedes and replacing the CEO in the case of BMW.
Most of these impacts come from Tesla’s Model 3 and the vehicle’s success in selling into the same compact luxury sedan segments that BMW, Mercedes and Audi have the most volume in. The Model 3 is something different and something entirely faster and dramatically technologically more advanced. For example, the Tesla Model 3 Performance was put up against the BMW M3 last year by Top Gear and beat it in terms of track time and braking, which is quite unexpected for the performance version of Tesla’s ‘mainstream’ electric vehicle. The performance of the vehicle is fairly important to brand reputation and sales, but what’s even more important is the experience that the customer feels and the automated capabilities of the vehicle. That’s where Tesla is miles ahead of the competition.
Many of Tesla’s competitors are still living in an era where a customer purchases a vehicle once and the only thing the manufacturer does is deal with warranty and recalls. On the other hand, with the initial introduction of centralized computing platforms based on NVIDIA, Tesla has effectively built a software-defined vehicle that is a constantly living and breathing platform evolving with the needs of users and Tesla’s roadmap over time. While Elon may have some lofty expectations for self-driving capability in Tesla vehicles, the company’s iterative approach to software is helping them inch ever closer to a self-driving vehicle beyond today’s Level2+ capabilities. The way that Tesla is able to accomplish this is through a multi-year effort to build their own high-performance custom computing architecture for FSD (Fully Self Driving) which includes multiple AI accelerators. Tesla is heavily using machine learning to accelerate what the car sees and how it can react to it while also adding features like Sentry Mode which give customers peace of mind. The system is not without its flaws, as the NTSB discovered in their investigation of a recent death in a Tesla where they blamed the company, CalTrans and the NHTSA.
Tesla’s FSD computer, which to date is still not capable of ‘Fully Self Driving’, claims a compute performance of 144 trillion operations per second, a significant increase from their use of NVIDIA’s Drive PX2 in the previous version of Tesla’s compute solution. Tesla’s new FSD computer has capabilities that have dumbfounded the other auto manufacturers to the degree that they aren’t really sure what to do. A Nikkei tear-down of the Tesla Model 3 found that Tesla’s electronics are 6 years ahead of the #1 and #2 car manufacturers in the world, Toyota and VW. They even said an engineer at a major Japanese auto manufacturer was stunned, saying that, “we cannot do it.” This puts many of Tesla’s competitors at a significant disadvantage in terms of keeping up with the company’s capabilities and how close they could be to fully autonomous driving capability. With Tesla already putting the hurt on them last year, the near future doesn’t look bright for them unless they adapt, quickly.
How do they escape?
Car manufacturers are scrambling to respond to Tesla, they have been for quite some time. Some are partnering with companies like Waymo, while others are buying their own autonomous solutions like Argo.ai and Cruise. However, many of these companies still do not have access to the same hardware, software and chips that Tesla does, and while some of them have some experience in software, it’s what I would call rudimentary, comparatively. It’s like the difference between cloud-native cloud and lift and shift of virtualized apps, a big difference.
To be able to do what Tesla did, you would have to hire some of the best architects in the world and keep them hidden from the world for years until they come up with a chip that works. Tesla effectively did that by hiring people like Intel’s Jim Keller and you can hear what he thinks about autonomous driving. As far as I can tell, none of the auto manufacturers have the people or willingness to spin up an entire chip division to build chips for their own needs.
Some manufacturers have tried to build entire new software architectures to compete with Tesla, but have run into speedbumps. Most recently, VW ran into some significant software issues with the VW ID3 which was slated to show up later this summer. However, the car’s software appears to have been too hastily developed and is suffering from significant bugs and errors as a result and may have to be delayed to be considered a shippable product. Software architectures, especially when built from the ground up are not something that should be approached lightly or can be done by simply throwing resources at the problem. These things take time and learning over many years of building these systems for electric cars and eventually they become more polished and ready for primetime.
There are a few manufacturers out there today that are offering solutions to auto manufacturers that can enable them to have a fighting chance against Tesla, especially the European manufacturers. Tesla has decided to go it alone when it comes to building their computer architecture, but Tesla’s competitors have options and that’s a good thing for them because there’s a chance that those suppliers could eventually surpass Tesla. Chip companies are very good at building the latest and greatest architectures, especially when they know that they have customers waiting for them. The auto industry has been slow to adopt the latest chip technologies from the leaders in the industry and it appears that indecisiveness is now really hurting them significantly.
Where do we go from here?
The auto manufacturers that have found themselves caught off guard are not completely sunk, even if they lack the expertise. Tesla is not the only company in the world manufacturing AI chips for use in automotive, specifically for autonomous driving capabilities. Manufacturers need to look long and hard at what their options are and consider that they cannot do this alone and that they need to partner closely with someone who has the experience and know-how to build these things and iterate them quickly. And since none of these auto manufacturers are going to go out and repeat what Tesla did, they really should consider looking at a company that has the experience, like NVIDIA. NVIDIA has been building automotive computing solutions since the beginning of the Tegra days, before the DRIVE platform and has been doing so for over a decade. NVIDIA has now built the complete end-to-end infrastructure as well as the processor architecture to enable their partners to be competitive with Tesla, especially with the introduction of the latest generation DRIVE AGX Orin platform which stands to give Tesla a run for their money. NVIDIA’s approach is also software-defined, which I believe the company has learned over the years is the right path towards full autonomy and how to compete with Tesla. Auto manufacturers struggling to keep up with Tesla should seriously consider giving NVIDIA another look if they want to have any hope of being able to stay relevant. The self-driving dark-horse, Qualcomm, should also be evaluated.