SAI Newsletter #8 - September 25, 2018
Updated: Jan 29, 2019
Happy belated 中秋节! China had Monday off for the Mid-Autumn festival so apologies if you were expecting this newsletter yesterday. We will move the newsletter back to Tuesday delivery next week.
A couple things that I focused on that I wanted to mention briefly here.
One was Eric Schmidt postulating that there could(will) be two internets in the future, one for China and one for the rest of the world. This is not so far-fetched and I discuss this in a bit more detail later in this newsletter.
Second, I attended the American Chamber of Commerce – China (Am Cham China) Tech and Innovation Conference in Beijing last week with Kai-Fu Lee as the keynote speaker.
As a fellow Carnegie Mellon alum, I think it’s great that he’s been able to carve out the role as ‘thinker’ for the space. He had some very good perspective on how the AI space has evolved and where he thinks it’s headed. He compared China and the U.S.’s abilities, differences, and strengths and is very optimistic for the future although I may take that with a grain of salt since he’s investing in a lot of companies in the space as well.
If there are any particular topics you’d like us to cover, feel free to send over your suggestion to email@example.com.
If you know of anyone who would like to sign up for this newsletter please have them visit: www.sinoautoinsights.com. Thanks for reading.
The Sino Auto Insights team
For those of us who live and/or have lived in China, the thought of 2 separate internets does not only seem possible, but probable.
As westerners living in a bubble, we do miss having access to western sites that are blocked in China but for many of the 1.5B Chinese people, most of whom have never used YouTube or Facebook, they aren’t missing a thing. There are equivalent apps that are by many accounts as useful and convenient as their western counterparts.
If we move this into the direction of autonomous vehicles, there will probably be at least two totally functioning but totally different transportation ‘systems’ as well.
I think it would be reasonable to assume that since the governments are run differently, how they would treat autonomous vehicles, personal and commercial, and what priorities they would set as the parameters for the system would also be thought about differently due to each government’s attitudes toward data, who can gather it, what they can do with it and ultimately who owns it.
This article explains why most AI companies put miles on their fleet vehicles in sunny NorCal (with the exception of foggy SF) and Arizona to collect their data.
From a system design standpoint, weather related scenarios that the AI needs to be prepared for, might only be considered edge cases when compared to the much larger number of non-weather related scenarios.
But one of the main reasons why AV companies need to be conservative and risk averse about weather related scenarios is due to the fact that they can significantly increase the odds of fatal and/or multiple car, accidents. In other words, the edge cases could be the most problematic.
If we’re to assume that millions of actual, non-simulated miles, need to be driven by each AI company’s fleet of vehicles in order to be safe enough to drive in sunny weather then, in addition to hardware needing to be advanced enough, the same ratio of miles will need to be driven in bad weather as well.
This could mean that these self-driving companies will likely launch in mostly sunny environments and that if weather becomes a factor, they may shut down their service until the bad weather has passed.
This article alludes to what I wrote about in a piece for the just launched AmCham China magazine ‘Business Now’ about how the tech startups still need their automotive OEM partners if they’re to be successful. I equated their relationship to being ‘frenemies.’ Will try to post the article next week so you can read for yourself.
This article is mostly a deeper look into Argo.ai but it also gives a brief history of the roots of the autonomous driving space and other pioneers that are leading the startups building the AI systems that will literally drive cars in the future.
Companies like Argo and Aurora who started later than GM and Waymo, if they’re going to put out a viable product to compete and in Argo’s case help Ford stay relevant, will need to think about approaching the challenge in a different way since they won’t be able to catch up by doing to same thing.
I have selfish reasons for including this article in the newsletter since I have a lot of ties to each of the places I used in the title.
Would love to see them all do well. I have more confidence in some more than others though.
It was recently announced that Waymo quietly opened an office and hired some staff in Shanghai a few months back and one can assume that since Tesla plans to begin manufacturing their cars in China as well at a $2B, yet to be built facility that they will also open and R&D center in China sometime in the near future.
I’ve spoken with a few of the other American AI startups and they’re all keen to better understand China and the opportunity that exists here.
Google is beginning to ramp up their hiring in China, maybe to relaunch search, so do these American companies know something that we don’t?
It’s anyone’s guess how long this trade war will last, but if China does not allow western AI companies to test their fleets in China, wouldn’t that be a reason for the U.S. govt. to limit the Chinese companies’ abilities to do the same in the U.S.?
One thing that the Chinese AI companies will have to deal with that the Western AI companies do not, at least for now, is the fact that one of their competitors is giving their IP away for free. All else being equal, how will they compete?
For most foreign OEMs they will not need to ‘move’ manufacturing over to China for domestic consumption. This is largely due to the fact that Ford came VERY late to the China party and still has a significant share of their vehicles sold in China imported from the U.S.
They’re adding capacity and flexibility into their manufacturing facilities here but that takes time. Time they do NOT have.
I am and have been a big fan of Bob Lutz. He’s a legend in the automotive industry and has worked for most of the big boys including BMW, GM, and Chrysler. He’s also been bearish on Tesla since its launch years ago.
When I discuss Tesla with people they tend to think I am not a fan and frankly that couldn’t be further from the truth. If I was in the market for a car, I would definitely consider and used Model S, X or brand-new Model 3.
My viewpoint has always been that Tesla does not make money and have NEVER sold a car without a subsidy. That and their stock price has kept them afloat for a very long time. They could’ve used that time to get their ‘house’ in order but if it’s not manufacturing, its logistics being the bottleneck that keeps Tesla from delivering vehicles to patiently waiting owners.
My other argument has always been that once the traditional OEMs focus on the EV segment, they will launch so many products and be willing to sell at a loss that it’ll make it tough for startups to compete. I still believe that’s going to be happening by 2020-21. There will be a global onslaught of vehicle introductions, so much so that there will be pricing pressure that will squeeze margins at the startups and make it even harder for them to eke out any margin on their vehicles.
With that said, I believe Tesla will be in a much better position than the startups that are coming online in the next few years.