What do you think the key competitiveness of Robotaxi will be in the future? Will aggregator be the key? Or what about the capabilities of autonomous driving technology?
If Tesla or another competitor (e.g. Zoox) were to successfully deploy a fleet of robotaxis to compete with Waymo, there would likely be little differentiation from the customer's perspective. Ultimately, passengers are primarily concerned with getting from point A to point B rather than brand. However, if more players were to enter the market, the underlying technology would likely be commoditised, with the majority of value captured by the aggregation platform. In a dual-player market though, the value would still predominantly accrue to the autonomous vehicle operators.
This was interesting Charlie. Having experienced both Waymo and FSD as a passenger, my sense is there is a DISTINCT possibility that success or failure is mostly up to approach. My conclusion on this is perhaps some many paths to autonomy will end as fools errands. Waymo offered their first driverless ride in late 2015. Tesla began offering FSD soon after in 2016. By late 2020 Waymo began offering paid rides to customers! The approaches could not be more different. If the key to learning how to autonomous driving is real miles Tesla should be all the way there as they have 1M+ vehicles using FSD while Waymo has reached a reliable scalable solution with at best 1000 vehicles. That's 1000X more raw data. How could the efforts of such different scales turn out the way they have? My sense of this is Alphabet from the start envisioned this as having NOTHING TO DO with miles driven. Instead they focused from the beginning on creation of a vivid simulation of driving from first principles. Every ride in every city is optimized to evaluate and generate edge cases. I am most interested in Waymo's entry into the Tokyo cab market. The largest taxi market in the free world 2.5X the population of NYC, narrow streets, unrivaled safety environment for pedestrians and right hand drive to boot.
I formerly wrote regularly on Substack and am thinking to return to the platform in the new year. If that is the case, I will consider subscribing to your Substack.
Thanks for reading Mark! I do agree that the fact that Tesla is nine years behind Waymo in launching a commercial robotaxi service demonstrates that data is only a small part of the puzzle. LiDAR is likely a big part of Waymo's success thus far and it will be interesting to see how Tesla's driverless service performs when it launches later this year, given that it relies exclusively on traditional cameras coupled with some clever parallax maths. I do find it intriguing that Tesla continues to pursue this strategy given that Waymo has such a compelling proof of concept with LiDAR.
I am retired but in my early career I worked for a scientific company. My work was heavily involved in thermodynamics and flow dynamics and simulation of physical systems based upon those first principles. It was infinitely easier than modeling the behavior of so many entities at once, any of which can behave chaotically. We even did camera systems for all sorts of unusual applications, often for the government.
Here are a few of the things on the technical side (the business model adds orders of magnitude to the complexity) which I believe have been critical to getting close to a commercial model. My sense is this has ever been part of the Tesla thesis for example (1) Small pool of vehicles (2) simulation to supplant the need for data (3) Precision-Mapping (4) low speed validation of the model (FireFly) (5) Instrumentation trial and error and ruggedization. I cannot imagine Waymo could be at the position they are today unless they did ALL 5 and this list is abbreviated
It will be great if Tesla progresses as competition is a wonderful thing. I just figure there are many other key strategies that Waymo embraced that will cause the same level of challenge. LiDAR is just an instrument and not worth arguing about. Most of the other issues are about philosophy and continuous improvement. Tesla has made many decisions that prevent continuous improvement of their solution. That mostly starts with imbedding unproven hardware on millions of vehicles on the bet that nothing will need to change. That was likely rooted in the opportunity to sell a lot of subscriptions which was certainly lucrative. It just made a whole host of good decisions no longer possible on the road to autonomy.
What do you think the key competitiveness of Robotaxi will be in the future? Will aggregator be the key? Or what about the capabilities of autonomous driving technology?
Hi Jeya,
If Tesla or another competitor (e.g. Zoox) were to successfully deploy a fleet of robotaxis to compete with Waymo, there would likely be little differentiation from the customer's perspective. Ultimately, passengers are primarily concerned with getting from point A to point B rather than brand. However, if more players were to enter the market, the underlying technology would likely be commoditised, with the majority of value captured by the aggregation platform. In a dual-player market though, the value would still predominantly accrue to the autonomous vehicle operators.
This was interesting Charlie. Having experienced both Waymo and FSD as a passenger, my sense is there is a DISTINCT possibility that success or failure is mostly up to approach. My conclusion on this is perhaps some many paths to autonomy will end as fools errands. Waymo offered their first driverless ride in late 2015. Tesla began offering FSD soon after in 2016. By late 2020 Waymo began offering paid rides to customers! The approaches could not be more different. If the key to learning how to autonomous driving is real miles Tesla should be all the way there as they have 1M+ vehicles using FSD while Waymo has reached a reliable scalable solution with at best 1000 vehicles. That's 1000X more raw data. How could the efforts of such different scales turn out the way they have? My sense of this is Alphabet from the start envisioned this as having NOTHING TO DO with miles driven. Instead they focused from the beginning on creation of a vivid simulation of driving from first principles. Every ride in every city is optimized to evaluate and generate edge cases. I am most interested in Waymo's entry into the Tokyo cab market. The largest taxi market in the free world 2.5X the population of NYC, narrow streets, unrivaled safety environment for pedestrians and right hand drive to boot.
I formerly wrote regularly on Substack and am thinking to return to the platform in the new year. If that is the case, I will consider subscribing to your Substack.
Thanks for reading Mark! I do agree that the fact that Tesla is nine years behind Waymo in launching a commercial robotaxi service demonstrates that data is only a small part of the puzzle. LiDAR is likely a big part of Waymo's success thus far and it will be interesting to see how Tesla's driverless service performs when it launches later this year, given that it relies exclusively on traditional cameras coupled with some clever parallax maths. I do find it intriguing that Tesla continues to pursue this strategy given that Waymo has such a compelling proof of concept with LiDAR.
I am retired but in my early career I worked for a scientific company. My work was heavily involved in thermodynamics and flow dynamics and simulation of physical systems based upon those first principles. It was infinitely easier than modeling the behavior of so many entities at once, any of which can behave chaotically. We even did camera systems for all sorts of unusual applications, often for the government.
Here are a few of the things on the technical side (the business model adds orders of magnitude to the complexity) which I believe have been critical to getting close to a commercial model. My sense is this has ever been part of the Tesla thesis for example (1) Small pool of vehicles (2) simulation to supplant the need for data (3) Precision-Mapping (4) low speed validation of the model (FireFly) (5) Instrumentation trial and error and ruggedization. I cannot imagine Waymo could be at the position they are today unless they did ALL 5 and this list is abbreviated
It will be great if Tesla progresses as competition is a wonderful thing. I just figure there are many other key strategies that Waymo embraced that will cause the same level of challenge. LiDAR is just an instrument and not worth arguing about. Most of the other issues are about philosophy and continuous improvement. Tesla has made many decisions that prevent continuous improvement of their solution. That mostly starts with imbedding unproven hardware on millions of vehicles on the bet that nothing will need to change. That was likely rooted in the opportunity to sell a lot of subscriptions which was certainly lucrative. It just made a whole host of good decisions no longer possible on the road to autonomy.