Uber’s Days Are Numbered: What Most Narratives Miss

Tesla’s bet to kill Uber is starting to pay off, and the reason is blunt: Uber does not own the cars that pick you up. The company lives by convincing people to bring their own vehicle and their own time when the fare looks good enough. That model worked for a decade. It starts to look fragile the second a factory can manufacture the labor Uber has always had to recruit.

Uber’s whole system runs on that recruitment loop. In the United States, about a million people is a fair picture of active drivers. Most of them are not full time. Roughly seven in ten drive part time, around 20 to 23 hours a week. Drivers only show up if the money is worth it. That is why you get surge. The app decides demand is hotter than supply, so it raises the price to lure more humans onto the road. Every busy night is a fresh negotiation with labor Uber does not control.

Tesla wants the driver built into the car. Open the Uber app and a person drives over because the fare looked good that day. Open Tesla’s robotaxi app and a car with nobody in the front seat pulls up for a paid trip. Both still get you across town. One depends on convincing a human to show up. The other depends on building a machine that can drive for hire without an hourly paycheck in the seat.

Uber is still the giant. Roughly 40 million trips a day go through the company. About 10 million people earn on the platform in a month, counting drivers and delivery workers. About $190 billion a year passes through as fares and fees. Almost all of that still rests on human driving labor. Driverless paid rides are still early. Waymo runs hundreds of thousands of paid autonomous trips a week across a few cities with a few thousand cars. Tesla’s paid robotaxi service went live in Austin in June 2025, started supervised rides there in early 2026, and expanded to Dallas, Houston, and Miami by mid year. The main car is still the Model Y. The fleet today is only tens to low hundreds of cars, not thousands. That is about to change as more Cybercabs ramp. Autonomous ride share is still roughly one tenth of one percent of global rideshare trips. Humans still rule the marketplace. The question is what happens if one company can manufacture the labor Uber recruits.

This is not chatbot AI. This is AI as continuous physical work inside a machine you build in a factory and then run for hours. That difference shows up first in the cost of a mile. Most of what you pay for a human ride today still pays for the person: time, energy, upkeep, depreciation, insurance, and the hassle of strangers in the back seat. A robotaxi still costs real money. The car wears out. It needs energy, cleaning, service, insurance, and sometimes a remote person to help. What drops out is the wage for every hour the car is online. Remove that wage from each mile and the lowest price a ride can support can fall if the cars stay busy enough. Long run models and company targets talk about running costs around 20 to 40 cents a mile once the system is fully scaled, versus Uber somewhere north of $2.80 per mile on average, sometimes a lot more.

Matching human labor with machines is a manufacturing problem. One hard working robotaxi that runs about 12 to 16 hours a day, seven days a week, can cover the online hours of about four to five average part time drivers. That means you only need roughly 200,000 robotaxis to match United States Uber online hours. You do not need to talk a million people into logging on. You need to build, approve, and keep busy that many cars that drive themselves. Tesla today builds about 1.6 to 1.8 million cars a year, and every one of them ships with FSD, the same self driving software running the robotaxi Model Y fleet. The company that can manufacture the labor at scale owns the game. To attract more drivers, you raise the price. To attract more self driving cars, you make them.

Once the cost of a mile can fall, it starts to matter who owns the hard parts of delivering the ride: the car, the software that drives it, service, booking, and matching. Tesla is trying to own all of it. Uber owns a different job. It owns the marketplace that matches riders to cars and the customer app millions of people already open without thinking. While a robotaxi fleet is still thin, that matching still helps. Empty seats cost money, and a big marketplace can fill more of them than a brand new app. So while the fleet is small, Uber can still be useful as the place where demand already lives.

But once Tesla has enough people booking through its own app, Uber is no longer the main way to fill the car. Tesla already started robotaxi rides on its own app, not on Uber. When that same car can already find riders there, putting the trip on Uber makes no sense. It adds cost. The rider pays more. Tesla keeps less. The marketplace takes a cut of a trip that could happen without a middleman. On Uber’s ride business, revenue has been about 30% of gross bookings in recent numbers from the end of 2025. That is a large bite if someone else already owns the car, the driving software, the service, and the customer. Once Tesla has its own riders, that money starts leaking away.

Price still decides a lot of what happens next. When cheaper machines offer rides for less than a human driver network can live on, the marketplace gets stuck between two bad choices. If the app keeps fares high enough that people still want to log on, the cars without drivers undercut on price and wait time, so bookings slide toward the machines. If the app cuts fares hard, human drivers take home less and many stop showing up. Once they leave, waits climb in the places the machine fleet does not yet cover: the edge of a city, late nights, a messy weekend rush. The network that used to feel thick starts to thin out. Human wages cannot compete with robot wages. If robot miles dominate, humans leave. If humans leave before the machine fleet is dense enough, your network gets smaller.

A mixed fleet can push that squeeze further. Put robotaxis and humans on the same app and the machines tend to grab the short, clean, easy trips that used to keep a driver busy and paid, while humans get more of the leftovers. Think of a store that installs self checkout and cuts too many cashiers before the machines can handle the busy rush. You get a longer line. The new supply can eat into the old supply that made the marketplace dense.

The network effect people credit Uber with is really local density. When more cars sit nearby, waits fall, more riders open the app, and those cars stay busier. That loop works while Uber still controls the largest pool of people willing to bring a car and their time. Without enough cars nearby, the wait stretches and the loop runs the other way. Driverless cars are going to drop the wage that loop depends on.

Today a lot of people open Uber because it is convenient at an okay price, not because they are loyal to the brand. That habit does not stick. If something shows up that is cheaper, faster to arrive, more private, and more consistent, people can switch without much drama. That is Uber’s biggest risk. And when a ride gets cheap enough, people take trips they skipped before. Older riders who do not drive. Households that can drop a second car. Trips that were never worth it because parking was a pain. Cheap manufactured rides can grow the whole pie while shrinking the need for a marketplace that lives by recruiting human drivers.

Uber talks people into driving. Tesla wants to build the driver into the car and sell you the mile. If that works at scale with people booking straight on the operator’s own app, the marketplace goes from running the game to watching it. This is a dire situation for Uber. The loudest signal may already be on the table: Uber has started lobbying local governments to slow down self driving rollouts in places like New Jersey. If a company that size is spending political capital to hold back the cars, you should pay attention.

Check the video here.