Why keep adding cities instead of scaling up in one city? Answer: they can't afford to scale up. So this is the only way they can show "progress" to investors.
In other news, Waymo recently gave a "driverless" ride
to a San Francisco journalist. It didn't go great. First, it wasn't driverless. Maybe because of fog? The chaperone didn't really say. And Waymo says Gen5 can handle fog. There were also phantom braking events and the car got so confused at a drop-off point the safety driver had to put it into manual mode for a few blocks.
You want to keep adding cities because you don't want your system to just be good in one city, but to experience all the intricacies of different cities. Adding different cities is also a litmus test on your ability to scale your software, infrastructure and logistics.
I will give you one thing. Google has a huge problem in failing to market new technological innovation in a brand new categories. We have examples in google fiber, project tango, google glass, DayDream, google stadia, etc.
But google is also great at copying, look at Android, Google Home, ARCore, Pixel, PixelBook (although now canceled) for example.
What I believe they should copy is partially how Cruise is handling deployments. Launching services at night (10pm - 6am) where there is less traffic and craziness is brilliant.
This will allow Cruise for example to scale faster and get to 50 cities quite easily. It will also challenge Cruise to progressively build up their infrastructure and logistics to handle 50 cities.
Ofcourse this requires that the foundational tech required is done. Aka the system can handle fundamental tasks at safety levels greater than a human and an ML factory to improve. For example: navigating around, stopping for and responding to pedestrians/cyclists. Stop for and responding to any and all static objects. And for general driving, it just has to be good to a certain performance miles that when you put artificial limits on it (10pm-6AM), that it boosts safety above human drivers.
For example a system that is developed to be a L4 highway system that can drive 90 mph, perform lane changes, handle on-ramp/off-ramp, lane merges, construction, freeway exit, traffic jam, accidents scene, all the DDT on the divided highway. But the current safety performance is 10,000 miles before a safety disengagement is required.
But by artificially limiting the system to a L3 single lane, requiring a vehicle in front and under 37 mph in clear weather, it automatically boosts the safety performance of the system to 100,000 miles (10x) for example.
The same is the case by Cruise developing the system to handle all driving, all times of day, all weather. But artificially limiting it during initial deployments.
Since I believe Waymo clearly has the superior tech, this is something i believe they CAN copy.
Theoretically, you could get to 50 cities in 5 years. The limit then becomes your logistics and infrastructure (safety drivers, warehouse, cleaning cars, charging cars, charging station, maintenance engineers, roadside assistance, etc).
Hiring more safety drivers is not the roadblock. For example, if Cruise/Waymo wanted a fleet of 1,000 cars in 50 cities.
10 cars used for autonomous driving, while 10 (20 during the day) cars used with safety drivers.
If they decided to hire 1000 safety drivers.
At 15 dollars per hour they would be spending $31,200,000 a year.
At 20 dollars per hour they would be spending $62,400,000 a year.
This is cheap (and won't happen all at once) compared to the 500 mill Cruise is spending every quarter. Tacking 8 mil to that with the high reward is really nothing.
If you noticed, Waymo put no deadline to their start of test to launch in LA. While SF put a 3 month deadline to their start of test and launch in Texas. Now if Cruise can launch in 5 or more cities per year...