Last year I made a case that Intel (the owner of Mobileye) was a better car company than Tesla. The premise of my thesis was that making car has become a commoditized business with largely marginal margins. I used the Aggregation Construct and Smiling Curve to explain that car makers will see long-term diminished value while aggregators (Uber, Taxify) and parts producers like Intel, Bosch and Invidia will rule. Largely, it could be like the old PC era where microprocessor makers like Intel generated huge value while those packaging PCs like Dell and HP diminished. Indeed, companies making cars today will become like those packaging PCs in boxes.
In this videocast, I explain why Intel is a better car company than Tesla. The car business has three core transitions: operator (man to self-driving). ownership (individual to fleet) and technology (internal combustion to electric). Because smartphone will make aggregation more efficient (hey Uber and Lyft), the utilization factor of cars will improve and in future the need of owning cars will drop. The implication is that all that will matter is when you need to move from one spot to another, you order a car – you do not need to OWN a car because the easy of summoning one will improve. Who will care for the car brand? The consequence will be huge: car manufacturers will see lesser value while aggregators (Uber, Lyft) and components makers (e.g, Intel, Bosch) will rule. It will play out like the old PC world where Intel made money while HP, Lenovo and Dell struggled selling PC boxes. The cars will be the PC boxes of the future.
Interestingly, Tesla is moving into the chip business. It would be replacing Nvidia chips with its own in-house chips for its cars. Doing that will move its value to the edge of the smiling curve, making sure that it generates more value for its investors. If Tesla does that very well, it could even license the technology to other car makers.
Tesla has been relying on chips from Nvidia to power its autopilot and self-driving car features, specifically Nvidia’s PX2 platform. But last week, Musk “let the cat out of the bag” about the carmaker’s own chip efforts .
Another really interesting problem may be what this development portends for the future revenues of chipmakers like Nvidia, Intel, and AMD that are counting on the nascent self-driving car market to demand a lot of their top-end, most expensive products. Musk said Tesla’s home grown solution was 10 times faster than Nvidia’s at the same cost. Nvidia, though, has announced its own next-gen chip for autonomous driving called Pegasus, also with 10 times better performance. Perhaps there are other advantages to Tesla’s chips as well, such as lower cost or energy consumption. In any event, Wall Street shrugged off the news and Nvidia shares were unchanged last week (and are still up 30% this year).
NB: As I explained on LinkedIn, this is a premium product.
… the most premium product in the world is making AI chips. Even Intel does not have that capability. There are about 2 companies today that do that. I do not believe it is commoditized. Do not confuse advanced AI chips that can power cars of the future with chips for PCs and smartwatches. Samsung plans to invest $20 billion over 3 years as part of its efforts to build such technologies. There is nothing commoditized about that category of chip.