Why did the driverless revolution stall?

Between 2015 and 2020, the automotive world believed in a collective dream: Driverless taxis would blanket cities, cars would drive themselves, and driving would become one of the 'lost professions.' This idea was sold first from Silicon Valley to investment funds, from funds to automotive giants, and then to city administrations. Conferences were buzzing with talk of 'full autonomy by 2025'; Google, GM, Tesla, Baidu, Uber… They all used the word 'revolution.'

The calendar now shows 2026. Sensors have become more powerful, cameras cheaper, artificial intelligence models more advanced, and processors faster. But the revolution did not arrive, and the streets did not change. Today, cars can see lanes, recognize traffic signs, perform emergency braking, and warn the driver. Many brands sell these driver-assistance systems in mass production. However, these do not eliminate the driver; they only reduce the driving burden. Full autonomy is an entirely different level, requiring a completely different vehicle architecture from the ground up:

- Redundant braking, steering, and power systems

- Combinations of Lidar + radar + camera

- Processing gigabytes of data per second

- Map and telemetry connectivity

- Remote monitoring and intervention centers

In today's technology, none of these are theoretical. This is exactly what Waymo did in Phoenix and Cruise did in San Francisco. But it was done in small areas, at low speeds, and under intense human supervision. This is where the problem begins: While the driverless revolution was being imagined, no one accounted for the real economics of cities.

The cost of making autonomy work at a city scale is extremely high. For a city to become suitable for autonomy, road markings must be standardized, signs must be clarified, maps must be constantly updated, and sensor blind spots must be addressed—and none of these are investments that win votes from the electorate. For a mayor, public transport, earthquake preparedness, housing, the environment, and traffic are much more urgent topics. Autonomy does not even make the top 10 on this list. In other words, Silicon Valley underestimated the urban economy.

Another wall is liability law. A legal, not technical, question locks the system: Who is at fault when an accident occurs?

- The hardware manufacturer?

- The software developer?

- The map provider?

- The fleet operator?

- The sensor manufacturer?

- The city administration?

In a vehicle without a driver, the answer to this question suddenly locks up the insurance and legal systems. No company wants to sign off on unlimited liability, no insurance company can easily price this risk, and no city administration wants to say, 'I will take responsibility for the accident.' For the first time in history, technology has turned an accident into a distributed chain of responsibility, and the legal infrastructure is not ready for this.

Another problem is this: Full autonomy is not even full autonomy right now. Today, fleets employ remote monitoring operators, telemetry analysts, data labelers, and safety drivers. In other words, we took the driver out of the car but put them in a control room. Moreover, because they are more qualified workers, they command higher salaries. Therefore, labor costs are not decreasing; they are increasing.

The robotaxi idea was simple: Remove the driver from the car › lower costs › everyone wins. In reality, the exact opposite happened. A human-driven Uber is cheaper. That is why the robotaxi model could not convince investors, city administrations, or the insurance system. Autonomy worked technically but could not be sustained economically.

The fate of the electric vehicle, although discussed during the same period, was different. On the EV side, battery costs fell, charging infrastructure was built, government incentives were provided, and production scaled up. As a result, it became commercialized. On the autonomous side, there were presentations, tests, and pilot programs, but marketization never arrived. In short, one of the two technologies of the same era won, while the other was delayed. This is not a technological defeat; it is an economic defeat.

It would not be accurate to say the story is completely over. It has just changed direction. Instead of fully autonomous driverless taxis in cities, intermediate models can be designed, such as trucks & logistics, port & mining site vehicles, and airport & campus shuttles. In these areas, traffic is simple, routes are fixed, the area of responsibility is clear, infrastructure is under control, and the business model is profitable. In other words, autonomy will win in controllable, narrow corridors, not in cities or on highways.

The picture we have arrived at today is clear:

- Technology is largely ready

- The economics did not work

- The law did not provide answers

- Insurance could not price it

- Cities did not allocate budgets

- Politics did not see it as a priority

That is why the driverless revolution stalled.

Now the real question is: For whom, where, and for what economy does a driverless vehicle make sense? Without clear answers to this question, autonomy will not come to our doorstep. Cars may be able to see the road, but cities are still not ready to bear the cost, risk, and responsibility of this technology.

And perhaps we will look back and say: The technology was ready, but the urban economy, law, and politics could not carry this revolution.