Monday, September 4, 2017

The following video clip should challenge cherished assumptions about Tesla, Silicon Valley, and self driving cars (it probably won't, but it should)

Before we go on, if you haven't, take a few moments and read this (we'll be quoting from it later but you really need to read the whole thing).




Obviously, we need to be careful about reading too much into this. There is a limit to how much you should take away from PR releases and tech reporters. There are, however, a few things we can conclude with some safety.

First, the system looks good. By most accounts it is the equal if not the superior of anything else on the road and, in one respect, it represents a significant step forward. Cadillac's system actually watches the driver, a considerably more sophisticated and  safer solution than Tesla and Mercedes Benz's nudge-the-wheel, and far more elegant than Google's just-give-up approach to the driver interface.

We can also conclude that that both the executives and the legal department of General Motors are fairly comfortable with this technology.When Tesla started rolling out its autonomous technology, there was always the possibility that Elon Musk was just betting big and playing reckless. That is certainly not the case here, which leads us to challenge one of the most cherished beliefs of the narrative.

The one thing everyone, from the trendiest tech blog to the New York Times, knew about autonomous vehicles was that the primary challenge was not technology and engineering, but the regulation and legislation.There would have to be massive and inevitably glacial changes in both of these areas before things could move forward. In other words, we couldn't get any cool stuff until those darn bureaucrats got out of our way. It was a narrative with almost archetypal appeal, but we can now say with some confidence it was never true. Multiple car companies are rolling out driverless options despite no great bureaucratic shift because bureaucrats were never a problem. Up until now, we didn't have driverless cars because the technology was not good enough.

Another cherished myth that is seriously threatened is that of the Silicon Valley Savior. It was often assumed that anything as wonderful as a car that drove itself would have to come from that most magical places. You couldn't expect anything wondrous and disruptive to come from a Detroit or a Stuttgart. Thus, the contributions of companies like Google and Tesla were given a wildly disproportionate amount of coverage. Even Apple, each managed to do nothing except prove that it wasn't up to the challenge of making cars, got listed as a major player. This was always a distortion. Silicon Valley never had a lock on the research or even a clear lead.

The deep-seated belief in a magical Silicon Valley populated by powerful chosen ones who could will great advances to being played a large role in the enormous valuation of Tesla. At the risk of oversimplifying, the reasoning went something like this: amazing things are about to happen with automobiles; this sort of thing comes out of Silicon Valley; the only Silicon Valley company/chosen one in position to do these things are Tesla and Elon Musk.

That said, there was a case to be made for Tesla reaping huge profits from autonomous vehicle technology. Here Goldfarb lays out the details:

So is Tesla the next Honda, or the next Webvan? If one believes that Tesla is worth more than Ford or GM, one better have a decent theory of how Tesla is making current capabilities obsolete, why incumbents cannot replicate these capabilities, and whether Tesla will earn good margins in a post-Detroit world. In short, when betting on disruption, one has to identify how this disruption will occur — not just intone the magic word. I identify four bets that Tesla is plausibly making: the EV bet, the autonomous vehicle bet, a software-platform-in-a-car bet, or the clean-energy-and-battery-company bet. None of these bets appear particularly promising.



The autonomous-vehicle bet

One aspect of the software strategy is so important that it deserves its own separate category. Perhaps Tesla will win the autonomous vehicle race. Alphabet (a.k.a. Google) is famously in this space, and Intel just bought the Israeli startup MobileEye for $15 billion. Uber raided Carnegie Mellon to begin to develop this capability, and every major manufacturer is investing in automation technologies. This is a very competitive space. But it’s undeniable that Tesla is at the forefront of self-driving technology that’s already on the roads. Tesla is able to learn as the technology is used — and this may be Tesla’s secret sauce: the difficult-to-get information that is necessary to perfect autonomous vehicles.

Driving is one of the most dangerous activities most people do on a regular basis. The technological problem of building software and sensors that replicate what a good driver does is difficult — and based on understanding rare events. (One Tesla driver, Joshua Brown, thought the technology was more advanced than it was and died when his auto-piloted Model S failed to recognize a semi-trailer and his car slid under it.)

Of all of the theories involving Tesla and disruption, this is the most intriguing — and convincing: Tesla is the first to roll out an autonomous driving technology. Its use allows Tesla to learn and improve, leading to a technological leadership position. Since the technology relies on machine learning, and machine learning only gets better with more data, Tesla is then able to develop a leadership position that becomes difficult to assail — analogous to Google’s in search. Tesla is then able to monetize this dominant market position.

The trouble, once again, is that at least 33 other companies are also developing autonomous vehicle technology, and also gathering a great deal of data. Any successful strategy would require a quick rollout that would prevent competitors from catching up. Mercedes’ new models already have advanced driver assistance capabilities similar to Tesla’s. Cadillac is expected to deploy within months. But perhaps the software platform is integral to this. Perhaps the software bet blends with the autonomous-vehicle bet! Tesla recently showed off this capability when it curtailed some of the autopilot features of Teslas with a software update.

The presupposition is that Tesla has and maintains an insurmountable lead. With the Cadillac generation of self driving technology, it is not clear that Tesla has any lead at all. Just to be clear, this does not mean that Tesla is out of the race or that it won't go on to a long and profitable future, but it makes it increasingly unlikely that these profits will reach the sky high level needed to justify what people have been paying for the stock.

No comments:

Post a Comment