GM to Have Robotaxi Service by 2019: Can Uber Survive?

GM claims it will have a fleet of robotaxis by 2019 and that will be its biggest profit-maker within a decade.

GM executives, speaking at an investor conference Thursday, said the company aims to run a large-scale fleet of driverless cars in big cities by 2019. GM is among the first major driverless-car developers to attach a timeline to the commercialization of autonomous vehicles, and the 109-year-old auto maker is racing big tech companies and Silicon Valley startups to lead the reinvention of the way people own and operate cars.

GM last year earned about a profit margin of 7.5% on its $166 billion in annual revenue from global car sales. Chief Financial Officer Chuck Stevens said that the company believes a driverless-car service by 2025 will offer 20% to 30% margins and a “total addressable market of several hundreds of billions of dollars.”

Ford Motor Co. has made similarly lofty margin projections related to mobility-related services in the past, but has offered far fewer details and stayed away from specific timetables.

Alphabet Inc.’s self-driving car effort, called Waymo, has outlined plans for a future autonomous-taxi service but hasn’t said when it would launch. It said this month that it would begin offering demonstration rides to members of the public without a safety operator behind the wheel, which goes against standard procedure in the industry.

Uber Technologies Inc. also has signaled plans for a sizable robotaxi fleet, saying it would order up to 24,000 Volvo SUVs to convert to autonomous vehicles in coming years, but it also hasn’t disclosed a timetable for commercial deployment.

Mr. Stevens said GM could offer rides for less than $1 a mile by 2025—down from around $2.50 for driver-based, ride-hailing services today. That could generate profit margins roughly quadruple the profitability of GM’s core car-making business, which generated $12.5 billion in operating profit last year.

By 2019?

I question 2019. It would require legislation that has not even begun. It would also require improvements to existing technology.

However, I have no doubts about a 2021-2022 timeframe.

It should be pretty clear at this point that all the deniers are delusional.

Can Uber Survive?

If GM, Ford, and Waymo offer taxi service, what exactly does Uber bring to the table and how can they compete?

Uber is in a huge legal dispute with Waymo (Google), a battle I expect Uber to lose. If so, it may have to pay royalties to Waymo, after purchasing cars from a manufacturer.

GM and Ford can obviously get cars cheaper than Uber.

OK. Uber has an app and drivers that use the app. But how long will it take for customers to switch to a GM, Ford, or Waymo app is someone else offers a better deal?

Mike "Mish" Shedlock

No. 1-15

There is not ONE fully autonomous car driving around our cities yet. Not ONE. They are only on controlled test tracts and carefully monitored by someone ready to grab the wheel if they F up. I think fully autonomous cars are totally OVERHYPED. The world has never seen anything do anything this complex without a human watching over it. Even trains which only run down tracks (INFINITELY less complex than navigating cities, highways, etc...) have a human in the cockpit. It is one thing for the Internet to invented, or web phones, or GPS. But a car driving by itself navigating every single curve, every traffic circle, 4 way stops with no lights, swerving out of the way of debis in the road, etc..... I say fully autonomous is 20 yrs away minimum. Just one of those causes a 20 car pile up and deaths and the technology will be set back 5 yrs. Who will get sued? Also what if a terrorist puts a bomb in a autonomous vehicle and sends it to a school right when all the kids are leaving for the day? How do you prevent that?


Cool, guaranteed to happen in two years. Knew it all along.


I should add that the Las Vegas autonomous bus failed within an hour, again due to something unexpected happening in the real world. (see ) It failed for the same reason that Cruse failed - it could not understand something going on in its environment, something that a human driver would have no trouble understanding and handling. This is the hard part of making autonomous vehicles. So far they have only done the easy stuff - sensors, mapping, etc. The hard part is understanding what is actually going on out there, understanding other road users' intentions, dealing with the unexpected and unpredictable messiness of the real world. There is a long way to go on this, as you will see.


The more astute observer will have noticed that Cruse's recent demo of their technology to journalists failed when something unexpected happened (Taco truck). And this would be on a demo router that they had mapped extremely carefully beforehand, that they had no doubt tested on extensively and thought they knew. And then on the day something unexpected happened, like it does in the real world.


We'll see. I think self driving cars will be somewhat commonplace in 10 years. 5 years is too soon. There are too many hurdles and government regulations. Being involved in software development my whole life, it takes 50% of the time to get 90% done. The last 10% takes forever. I'm not even sure we're at the 90% done mark yet.