What the future looks like
Most analysis asks one question: what share of Uber, Lyft and taxi spending can driverless cars take? We think that frames the opportunity far too narrowly. The driver isn't just a cost line. The driver is the reason a car sleeps, commutes, refuses the 4 a.m. airport run and sits idle in a parking space. Take the person out and the vehicle becomes a different kind of asset: one that runs around the clock, moves itself to wherever demand is about to appear, and can do a different job at different hours.
Picture a single vehicle's day five to ten years out. Early morning it stages at the airport. Then it carries commuters, then takes an older passenger to a clinic, then delivers restaurant orders, then runs a sightseeing loop, then works the late-night crowd leaving a stadium. Nobody needed to drive it between any of those jobs.
That is why we see robotaxis as the start of a new mobility economy, not just a cheaper taxi.
Five layers of the robotaxi opportunity
1. Replacing ride-hailing. The obvious starting point, and where almost all of the market's attention sits today: trips that now go to taxis, Uber, Lyft and private hire.
2. Replacing car ownership. The bigger competitor may be the second car in the driveway. Many households own two vehicles only because two people occasionally need one at the same time. If a car reliably shows up in minutes, the case for paying for a second vehicle's insurance, upkeep, depreciation and parking gets weak. The comparison shifts from robotaxi versus Uber to robotaxi versus owning another car.
3. New riders and commercial mobility. Children, teenagers too young to drive, older people who have given up their keys and people with disabilities could all travel on their own in a properly regulated system. That is new demand, not stolen demand. Businesses join too: hotels, airlines (crew transport), hospitals (non-emergency patient trips), factories (night-shift transport), campuses and resorts could contract fleets directly. Airports suit fleets especially well, with concentrated demand, long trips and odd-hour arrivals, and they put robotaxis up against rental cars and airport parking as well as taxis.
4. Logistics. A vehicle doesn't need a passenger. In quieter hours the same fleet can move groceries, prescriptions, parcels, luggage and urgent business-to-business items such as lab samples or spare parts.
5. Platform economics. Once the platform knows where you want to go, it sits right before the purchase. Ask it for a good sushi place and it can recommend, reserve and drive you there. Add subscriptions, advertising, referral fees, guided city and wine-country tours, and mobility bundled into hotel stays or apartment rents (fewer garages, more included miles), and a fleet starts to look like a commerce layer.
Under all five layers sits a fleet that can position itself before demand shows up, for example moving cars toward an arena before the concert ends, and that can drive itself to charging and maintenance. Cabins designed for passengers rather than drivers also turn travel time into usable time, a source of value that fare comparisons ignore.
Our conclusion: valuing robotaxis by today's ride-hailing spend asks the wrong question. The better question is what round-the-clock, self-positioning physical infrastructure is worth.
How big could it get?
- $300–400 billion of passenger-car revenue from driver-assist and autonomous-driving systems by 2035 (McKinsey, 2023). This counts the technology sold with cars, not the service layers above.
- Robotaxis in the US alone: about $7 billion of annual revenue by 2030 from roughly 35,000 vehicles, around 8% of the US rideshare market (Goldman Sachs Research, 2025). That is layer 1 only.
- Autonomous trucks could haul about $18 billion of US freight in 2030, out of a roughly $660 billion market (Goldman Sachs Research, 2025), an early signal for the logistics layer.
- We found no credible published estimate for layers 2 to 5 (ownership replacement, commercial mobility, logistics with passenger fleets, and platform revenue). That gap is the point of our thesis, and we won't put a number on it until there's a sound source.
- A sobering counterpoint: McKinsey expects only about 1% of new vehicles sold in 2035 to be capable of Level 4 driving or above (McKinsey, 2025), so fleets, not private cars, are likely to lead.
Figures are third-party estimates, quoted as published. Different studies use different scopes and years, so they aren't always comparable.
What we're watching
- New cities, larger fleets and removal of safety drivers or monitors.
- Tesla's robotaxi service, which began as a small paid pilot in Austin in June 2025 (Reuters, 2025): fleet size, coverage area and supervision.
- Airport access agreements and business fleet contracts (hotels, hospitals, employers).
- Rules that let minors or unaccompanied older and disabled riders use driverless vehicles.
- Fleets mixing passenger trips with deliveries, and revenue from subscriptions, ads or bookings.
- Cost per mile: Goldman Sachs expects per-mile depreciation and insurance costs for robotaxis to keep falling as hardware gets cheaper (Goldman Sachs Research, 2025).
What could go wrong
- A single high-profile crash can pause a program or a city's permits.
- Timelines have slipped for a decade; promised dates are not evidence.
- Layers 2 to 5 are our thesis, not proven markets; regulators may limit unaccompanied riders, deliveries or in-car commerce.
- Fleet economics: vehicles, remote operators, insurance and depots are expensive, and margins are unproven at scale.
- Competition from well-funded operators in the US and China could compress pricing.
What the future makers say
“The autonomous vehicle revolution is here.”
“It is important to note that Tesla is by far the best in the world at real-world AI.”
Quotes are reproduced from the linked public sources; paraphrases are labeled. Inclusion doesn't imply any affiliation with or endorsement by these people or companies.
Companies building it
Only companies we own. Tickers only, never position sizes.
A fleet-learning approach to self-driving across its vehicles, plus its own robotaxi service.
Trains driving models in the data center and sells DRIVE in-car computers; Toyota is building next-generation vehicles on NVIDIA DRIVE AGX (NVIDIA, 2025).
Owns Zoox, which is building purpose-built robotaxis, and runs one of the largest delivery networks, a natural fit for the logistics layer.
Memory and storage for in-car computers and for the data centers that train driving models.
Sources
- McKinsey, 2023
- Goldman Sachs Research, 2025
- McKinsey, 2025
- Reuters, 2025
- NVIDIA Blog, Jan 2025
- CNBC, Sep 2025
Last reviewed: 22 Oct 2026.
Not investment advice. This page is opinion and education, researched and presented with the help of AI analyst personas, and may contain errors. The EarlyConviction team holds positions in TSLA, NVDA, AMZN, MU and may buy or sell at any time. Read the disclaimer.