Uber is currently attempting to win a race it already forfeited years ago. After burning billions of dollars on a homegrown self-driving division that ended in a tragic pedestrian death and a massive trade-secret lawsuit, the company pivotally shifted from builder to broker. Now, CEO Dara Khosrowshahi is trying to convince the world that his platform is the indispensable middleman for a robotaxi industry that doesn't yet exist at scale. The primary challenge is that Uber’s survival depends on companies like Waymo and Tesla needing a third-party app to find customers, a premise that looks increasingly shaky as the technology matures.
The math of the ride-hailing business has always been ugly. For over a decade, Uber stayed afloat by subsidizing rides with venture capital, essentially paying people to take cars they couldn't afford. When the "easy money" era ended, the company finally turned a profit by squeezing drivers and raising prices. But drivers remain the biggest expense, accounting for roughly 70% to 80% of the gross booking value. Removing the human from the driver's seat is the only way the unit economics of ride-sharing ever truly make sense.
The Platform Fallacy
Uber’s current strategy relies on the idea of being "AV agnostic." They want to be the Amazon of autonomous vehicles (AVs), providing the interface, the routing, and the customer base for anyone with a fleet of driverless cars. Recently, they announced a multi-year partnership with Waymo to offer autonomous rides in Phoenix and Austin. On paper, it looks like a win. Uber gets to offer "cool" tech without the R&D costs, and Waymo gets access to Uber’s massive user base.
But look closer at the incentives. Waymo, owned by Alphabet, has its own app. It has its own mapping data. It has its own fleet management. Why would a company that has spent upwards of $30 billion developing the world’s most advanced "Driver" give away a significant percentage of its revenue to Uber in the long run? In the early stages of a rollout, Waymo needs Uber for demand density. Once the fleet is large enough and the public is comfortable with the tech, the middleman becomes a parasitic expense.
The tech giants aren't the only threat. Tesla’s long-promised "Cybercab" represents a different kind of vertical integration. If Elon Musk succeeds in launching a dedicated robotaxi network, he won't be listing those cars on Uber. He will run them on the Tesla Network. Uber is effectively trapped in a pincer movement between Google’s software superiority and Tesla’s manufacturing scale.
The Ghost of Advanced Technologies Group
To understand the desperation, you have to look back at the wreckage of ATG (Advanced Technologies Group). Uber’s internal self-driving unit was once the crown jewel of the company, valued at $7.25 billion during a 2019 investment round. It was meant to be the insurance policy against a future where they no longer controlled the labor.
The project collapsed under the weight of hubris. The 2018 crash in Tempe, Arizona, where an Uber test vehicle killed Elaine Herzberg, didn't just cause a legal nightmare; it shattered the company's internal culture and moral authority. When Uber eventually offloaded ATG to Aurora Innovation in 2020, they didn't just sell a division. They sold their autonomy.
Now, Uber is a software layer sitting on top of an asset-heavy world. They don't own the cars, and they don't own the brains driving them. They own the "network effect," which is a powerful but notoriously fragile moat. If a rider can get a Waymo for $15 on the Waymo app or $22 on the Uber app (because of Uber’s take-rate), the network effect evaporates.
The Maintenance Mirage
Industry analysts often point to fleet management as Uber’s secret weapon. The argument goes that someone needs to clean the vomit out of the cars, swap the tires, and manage the charging cycles. Uber claims its experience managing millions of independent contractors translates to managing robotaxis.
This is a fundamental misunderstanding of the shift from a "labor" economy to a "capital" economy. Managing a fleet of autonomous Chevy Bolts or Waymo Jaguars isn't like managing a fleet of gig workers. It requires massive centralized depots, specialized sensors, and high-voltage charging infrastructure. Uber’s current business model is "asset-light." Moving into fleet management would require them to become "asset-heavy," the very thing they spent a decade avoiding to keep their stock price high.
Look at the numbers from existing pilot programs. A human driver provides their own car, pays their own insurance, and handles their own maintenance. In a robotaxi world, the fleet owner takes on all those costs. For Uber to maintain its current margins while paying for the hardware, the efficiency gains from removing the driver must be astronomical.
Global Fragmentation and the Regulatory Wall
While the US market is a three-way fight between Uber, Waymo, and Tesla, the global picture is even more complex. In China, companies like Baidu (with Apollo Go) and Pony.ai are already clocking millions of autonomous miles. In these markets, Uber is non-existent, having sold its operations to Didi years ago.
This creates a fragmented map where Uber has to negotiate city by city, country by country, with different tech providers. Every time a new autonomous player enters a market, Uber has to hope that player is willing to share the pie. If a local government in London or Paris decides to favor a homegrown autonomous shuttle service, Uber has no "driver" to offer and no "tech" to pivot to.
The Labor Counter-Strike
There is also the human element that the "robotaxi race" narrative tends to ignore. Uber currently relies on approximately 7 million drivers globally. As the company leans harder into autonomous partnerships, the friction with its human workforce will reach a breaking point.
We are already seeing "deactivation" protests and strikes over falling wages. If Uber starts actively prioritizing robotaxis over human drivers in high-demand zones, the resulting labor unrest could lead to crippling regulations. Governments are already looking for reasons to reclassify gig workers as employees. A company that tries to replace its "partners" with robots while still claiming those partners aren't employees is walking into a legal buzzsaw.
The Survival of the Useful
Uber’s best hope isn't in being a transportation company, but in being a logistics layer. They are betting that the complexity of moving people and things is so high that no single hardware manufacturer will want to deal with the customer service, the edge cases, and the insurance headaches.
But betting on the "laziness" of your competitors is a dangerous strategy when those competitors are Google and Tesla. These are companies that take pride in owning the entire stack.
The question isn't whether robotaxis will happen—they are already here. The question is whether a company that doesn't build the cars and doesn't build the AI can survive on a 25% commission for very much longer. History suggests that when a technology becomes a utility, the middleman is the first one to be cut out. Uber isn't fighting to stay in the race; it's fighting to prove the race still needs a referee.
Check the pricing the next time you are in Phoenix. If the Waymo app is cheaper than the Uber app, you are looking at the beginning of the end for the ride-share giant's dominance.
