If Robotaxis Get This Cheap, Why Own a Car?
(4min read) Our expert predicts they'll be 4x cheaper in 4 years.
Welcome to CoMotion Deep Dive, a weekly examination of an issue or technology shaping the future of mobility. If this email was forwarded, you can sign up here.
It’s hard to look at the success of Waymo, Tesla and other companies operating robotaxis and not believe that driverless cars will soon be the norm rather than the exception. I use “soon” flexibly –– it could be five years or 20 years.
What is far less clear, however, is whether driverless cars will fulfill what I’ve always considered their most exciting promise: allowing more people to abandon car ownership altogether.
Will the absence of a driver make ride-hailing so cheap that it no longer makes sense to buy a personal vehicle, even in a place like Dallas or Phoenix, where car-free living is almost nonexistent? And will it translate into meaningful cost savings for people who currently spend an average of $12,000 a year to own, maintain, fuel and insure a depreciating asset?
To answer that question, I talked with our own Jerry He, Head of Innovation & APAC Partnerships at CoMotion GLOBAL and Executive Director of the Corporation for Automated Road Transportation Safety (CARTS), a Princeton and New York-based nonprofit think tank.
To be clear, Jerry’s not the kind of guy to make sweeping, vibes-based predictions à la Elon Musk. He is a sober-minded data wonk. So I was stunned when he told me he believes that by the end of this decade, robotaxis will not only be widely available in the U.S., but they’ll only cost about a quarter as much as an Uber or Lyft today.
“I think in the grand scheme of things, we’ll shift towards a Mobility as a Service model,” he says. “What’s keeping society from achieving the car-less vision that Uber and Lyft promised 10 years ago is the high cost of labor and insurance.”
According to Jerry’s research, today you can expect to pay $2-3 per mile for Uber’s standard service, UberX, depending on geography (urban/suburban) and excluding tips. He believes that within the next 2-3 years, robotaxis will be able to achieve profits at only $1 per mile. Within 4-5 years, they’ll be able to operate profitably while only charging 50¢ per mile.
The evidence, he says, is what Tesla is doing in Austin. Musk’s many detractors have been eager to highlight errors made by his company’s small fleet of robotaxis, but overall the performance has been pretty good. Moreover, Tesla is delivering a strong performance at a lower cost than Waymo, the Alphabet-owned company that has by far the most extensive and successful commercial robotaxis on the road in the U.S.
Tesla, which has forgone LiDAR entirely, only spends about $60,000 per vehicle on hardware. That’s far lower than the six-figure sums other robotaxi providers have been spending for hardware.
Musk’s vision of turning every Tesla owner into a potential robotaxi operator is likely far-fetched, but the fact that Tesla is already mass-producing the vehicles it intends to use for its autonomous fleets gives it a major leg up over other companies trying to compete in the robotaxi space.
Jerry breaks down the cost of robotaxis into four parts:
Hardware –– vehicle, sensors, camera, EV chargers, etc.
Electricity –– to charge the fleets
Intelligence –– both human and artificial. In other words, the staff overseeing the robotaxi fleets and the AI software powering them.
Insurance – often overlooked by the end consumer, rideshare insurance costs can reach as high as 30%+ of total trip cost in certain geographies such as CA and NJ)
Self-driving cars will 'dramatically change' auto insurance
It is those last two costs that he expects to come down significantly in the coming years. The cost of intelligence “will ideally go down to close to zero since the marginal cost of software is approximately 0,” he says, while the cost of insurance will plummet as it becomes clear that autonomous vehicles are in fact much safer than human-operated ones. The property casualty insurance industry will eventually shift its risk profiling from “human operator errors” to “product liability”.
The one thing that Jerry’s model doesn’t account for is culture and politics. Even if robotaxis prove far safer than human drivers, it’s still likely that in the early years robotaxi accidents will prove more controversial and keep political leaders from fully embracing autonomous vehicles. Society is strangely at peace with tens of thousands of people dying each year on American roads –– as long as their deaths were caused by drivers.
But if Jerry’s analysis proves right, the question won’t be whether Americans can give up their cars — it’ll be why they’re still clinging to them.
Unlocking the New Mobility Playbook
Join us at CoMotion LA ‘25 on Nov. 12-13 and connect with the companies building tomorrow’s transportation, the investors backing them, and the city leaders making it happen.
Attend two days of stellar programming, engaging panels, workshops, demos, networking, energizing talks and after-hours gatherings, with a special OMF Summit Track that highlights the latest in open, transparent civic data.
CoMotion's mobility goodness brought to you by:
Jack Craver,
Editor, CoMotion NEWS
jcraver@comotionglobal.com
Write to us if you would like to suggest content for more deep dives!
Great piece. In my experience, much-hyped tech advances come slower than predicted and yet are more disruptive.
I am always amused by the notion that a strong move to AVs will somehow reduce car usage and dependency. Last time I checked an AV is a car, and if we all used "rental" AVs instead of current ownership-based car usage model, there would be no reduction in automobile VMT, there would in fact be an increase, perhaps very substantial (given 40%+ deadheading for TNC trips) due to non-AV owners needing to summon an AV in order to travel. Moreover, I suspect that a substantial fraction of vehicle owners would opt for purchasing an AV if the cost is reduced to the levels implied by this post, so no reduction in VMT in that scenario. In fact, that is the entire business plan behind Musk's robo-taxi initiative, that automobile owners will purchase a Tesla that can be rented out as a robo-taxi for many hours of the day/week and they will reserve it for personal use when they need it. In that scenario, we end up with fewer vehicles than today, but no reduction in VMT since all current automobile trips remain in automobiles. The simple reality is that for most trips in most urban areas--and for virtually all trips in all non-urban areas--the decision to use a car for the trip is the utility maximizing choice for the trip maker, and AVs never reduce the utility of such personal choices, and in some circumstances they enhance that utility. It is certainly interesting to know that this contributor predicts that AVs will REDUCE the cost of automobile travel in certain urban trip making scenarios, and it is clearly the case that for non-urban travel many/most motorists would prefer to let the AV technology control the vehicle. So while AVs will be disruptive in certain ways, they will more generally reinforce the dominance of automobile travel in our urban areas.