GM/Cruise CEO Dan Ammann announced today that Cruise would not meet its goal of deploying in 2019.
Cruise prototype self-driving car
Cruise
Nobody is particularly surprised by this, particularly in the light of leaks which suggested Cruise was seriously behind on milestones. In addition, rumors have swirled from departing Cruise employees of trouble within the organization. On the other hand, Cruise has recently been growing greatly, and added major investment from Honda, T. Rowe Price and Softbank totaling over $7B — a whopping amount.
Cruise has often touted, in an effort to compare itself with industry leader Waymo (where this author worked 7 years ago,) that it is testing and planning to deploy in the much more complex environment of San Francisco, not the easy streets of suburban Phoenix where Waymo One was launched, and many other companies test. Ammann says it’s 40 times harder. And they’re right — SF is much harder, and going after the harder problem will, eventually, make their car better and safer than just trying the easier streets. They’re right, but it’s a gamble. If you go after a problem that’s too hard, you will take longer. While you’re taking longer, others who took the easier approach will be deploying. Driving difficult streets teaches you important lessons, but deploying, and driving at much higher volumes, with real customers, also teaches important lessons. Testing with two safety drivers is expensive. The team that can deploy in an easier city with no drivers can operate more.
It’s not obvious which is the winning strategy. There is debate about the value of lots and lots of less interesting road miles. When it comes to testing, you are out trying to find “corner cases,” and for now, robocar teams are finding that corner cases also frequently occur in the middle of the block, too. (Same thing with edge cases.) If it’s really 40 times harder (ie. would take 40 times as long) then the Cruise bet is clearly the wrong one, since Waymo has taken several years to get where it is. In reality, teams get exponentially better as time goes on, so 40x harder may only mean taking 2-3 times as long, depending on your exponent. But it’s still a large gap.
Conversely, once you are in production on easy streets, it frees your test team to now focus on the next target, those complex streets. That approach will get to the difficult streets later, but operate at a larger scale. If you’re getting exponentially better through other means, then you can gobble up that “40x harder problem” much more quickly than those who started it soon.
The battle for control of the future of automotive may not involve head to head competition for some time to come. The competing teams, once ready to deploy, will probably move to untapped cities where no competitor is present rather than deliberately join the battle in a city that already has a player. Deploying in a city is hugely expensive, in capital and human bandwidth, so you have to do it one city at a time, even if you are as rich as Croesus (or Google or Apple which are richer than Croesus.) Deploying first in a city offers the promise of a big first-mover advantage there — or so everybody hopes. There are plenty of relatively easy cities to start with. San Francisco is special, though. It’s home town to a large fraction of the teams and the high tech world.
The press this year has been full of stories about the arrival of the “trough of disappointment” on Gartner’s famous hype cycle. While some companies were fairly low-key in their hype (including Zoox, Apple, some of the car companies and yes, Waymo) others, notably Tesla were not. It’s easy to get into a hype battle, but every serious player has always known that this is not a problem that will get solved one city at a time, one street at a time. A lot of the disappointment comes from the fact that teams will brag about when they might pull off their first goals — limited service in a limited area — and the public takes it to mean a promise of complete service everywhere. (In addition, Elon Musk is the worst sinner here, having promised full service on all streets for times that are now moving into the past, and continuing to make such promises.)
