The Waymo Question: What Are We Actually Building All This For?

The promise of autonomous vehicles has always sounded grand: safer streets, cheaper rides, more access, fewer deaths, less congestion, and a future where transportation is available to anyone at the tap of a button.

But after more than a decade of development and billions of dollars invested, it is worth asking a more uncomfortable question:

What if the entire robotaxi vision is not actually about democratizing transportation? What if it is about turning transportation from something many people own into a premium, metered, corporate service?

That distinction matters.

Because if Waymo and similar autonomous vehicle companies cannot provide rides that are materially cheaper than Uber, Lyft, taxis, or personal car ownership, then the argument starts to fall apart. And if the safety gains are real but not uniquely achievable through robotaxis, then “safety” becomes less of a decisive reason to rebuild urban transportation around privately owned autonomous fleets.

At that point, the whole project begins to look less like a revolution in mobility and more like an expensive attempt to insert another corporate margin into a basic human need.

The unit economics are still the problem

The simple robotaxi pitch is that removing the human driver removes the largest cost in ride-hailing.

In theory, that sounds compelling. Uber and Lyft rides are expensive partly because a human being has to be paid. Remove the driver, and perhaps the service becomes cheaper.

But that is only one side of the ledger.

A Waymo vehicle is not just a normal car without a driver. It is an expensive electric vehicle loaded with sensors, cameras, lidar, radar, compute hardware, connectivity systems, and redundant safety infrastructure. It requires depot operations, charging, cleaning, maintenance, software support, mapping, monitoring, remote assistance, roadside support, insurance, and periodic intervention when the vehicle gets confused or stuck.

Each car is not just a car. It is a rolling data center, fleet asset, robotics platform, and liability exposure.

And the AI stack is not a one-time fixed cost. Every new city means new roads, weather patterns, construction zones, edge cases, local driving behaviors, emergency vehicle protocols, school zones, pickup/drop-off issues, regulators, maps, and validation work. Expansion means more training, more simulation, more inference, more monitoring, and more engineering support.

So the question is not simply, “Can a Waymo car drive without a driver?”

The question is:

Can a Waymo car drive without a driver cheaply enough to beat the human alternative after accounting for the full operating cost of the system?

So far, that remains unproven.

In fact, current public price comparisons suggest Waymo is still more expensive than Uber or Lyft in at least some markets. A 2025 Obi analysis cited by TechCrunch found average Waymo rides at $20.43, compared with $15.58 for Uber and $14.44 for Lyft. Later reporting suggested the gap may be narrowing, but the core point remains: the robotaxi is not yet obviously cheaper than the human-driven ride.

That is a big problem for the entire thesis.

If the robotaxi is more expensive than Uber, then the absence of a driver did not actually make transportation cheaper for the rider. It just moved the cost somewhere else: into sensors, depreciation, AI teams, fleet operations, remote assistance, capital costs, and corporate overhead.

The “safety” argument is not enough

The strongest argument for autonomous vehicles is safety.

To be fair, Waymo does have safety data that deserves to be taken seriously. A 2025 study comparing Waymo rider-only crash rates across 56.7 million miles found statistically significant reductions in several serious crash categories relative to human benchmarks, including injury-reported and airbag-deployment crashes.

That is meaningful. It should not be dismissed.

But it still does not settle the broader policy question.

First, the comparison is usually between Waymo and broad human-driving benchmarks. That is not necessarily the right comparison. The true substitute for many Waymo rides is not “the average distracted, tired, angry, underinsured, texting consumer driver.” The substitute is often Uber, Lyft, taxis, airport shuttles, public transit, professional chauffeurs, or other managed transportation services.

A professional driver is already different from a typical private driver. They drive for a living, are usually screened, are subject to platform rules, have ratings, and often drive newer vehicles. Uber and Lyft are far from perfect, and rideshare work creates real fatigue and safety risks. A 2024 study found rideshare drivers can face elevated crash risk due to factors like fatigue, phone use, and unfamiliar roads.

But that cuts both ways. If the public goal is safer paid transportation, we do not need to jump immediately to a fully autonomous robotics fleet. We could regulate and professionalize ride-hailing more seriously: better driver screening, better pay to reduce fatigue, stricter hours-of-service limits, better telematics, safer vehicles, better insurance, and higher operating standards.

Second, the safety gains from autonomy are not the only available safety gains. The National Highway Traffic Safety Administration already has powerful tools to reduce fatalities across the entire vehicle fleet. For example, NHTSA finalized a rule requiring automatic emergency braking, including pedestrian automatic emergency braking, on new passenger cars and light trucks starting in 2029.

That matters because broad safety mandates apply to millions of vehicles, not just a limited robotaxi fleet in a handful of geofenced cities.

If the goal is to reduce deaths, regulators can require more of the basic technology that already works: automatic emergency braking, pedestrian detection, lane-keeping assistance, blind spot monitoring, better traction control, driver monitoring, improved headlights, stronger crash structures, more airbags, speed assistance, and safer vehicle designs for pedestrians and cyclists.

NHTSA itself says driver-assistance technologies have the potential to reduce crashes and save thousands of lives each year. It also reports that 39,254 people died on U.S. roads in 2024.

That is the scale of the problem. And that scale raises a serious question:

If we want to save lives, why would the main solution be a small fleet of expensive robotaxis instead of mandatory safety improvements across the entire national vehicle fleet?

Autonomous vehicles may eventually be safer than human drivers in many contexts. But “safer than average human driving” is not the same as “the best, cheapest, fastest, or most democratic way to improve transportation safety.”

Turning car ownership into a metered service may make life more expensive

There is also a deeper economic issue.

For many households, owning a car is expensive. Insurance, repairs, gas or charging, parking, registration, depreciation, and financing all add up. But ownership also has one major advantage: once you own the vehicle, the marginal cost of each trip can be relatively low.

A family that owns a used Toyota can make a grocery run, school pickup, doctor visit, late-night pharmacy trip, or weekend drive without paying a platform margin every single time.

Robotaxi economics invert that.

Instead of owning an asset, the person becomes a perpetual renter of transportation. Every trip becomes a transaction. Every mile becomes billable. Every pickup includes not only the cost of the vehicle and energy, but also fleet overhead, software costs, insurance, capital recovery, investor return, and profit margin.

That may work for some urban professionals who already use Uber frequently and do not want to own a car. It may also work for tourists, business travelers, people with disabilities, people who cannot drive, or households that only occasionally need a vehicle.

But it is not automatically democratizing.

For many people, it could mean taking one of the largest personal expenses in life and making it even more financialized. Instead of paying for a car, people pay forever for access to a transportation network owned by a giant corporation.

That is not liberation. That is dependency.

And it becomes especially questionable if the service costs more than Uber or Lyft today, and far more than the marginal cost of driving a personally owned vehicle.

The labor argument is also uncomfortable

There is another awkward reality: the robotaxi model does not eliminate labor. It changes which labor gets paid.

Instead of paying drivers, the system pays AI engineers, robotics engineers, fleet technicians, safety operators, remote assistance teams, data labelers, mapping teams, hardware teams, policy teams, lawyers, managers, and executives.

Some of those jobs are valuable. But the economics are strange.

If replacing a $25-per-hour driver requires a massive organization of highly compensated AI workers, expensive hardware, specialized depots, remote support, and billions in capital investment, then society should ask whether this is actually an efficiency gain or just a prestige technology project.

That does not mean AI engineers are bad people. It means the business model has to justify their cost.

At some point, if the system matures, the engineering headcount and compensation structure would need to look very different. A mature transportation utility cannot carry the cost structure of a frontier AI lab forever. If it does, riders will pay for it through higher fares.

And that is the core issue: the customer eventually pays.

If the fares are high enough to fund the fleet, the engineers, the compute, the hardware, the maintenance, the insurance, the remote operations, the lobbying, the corporate overhead, and the return on billions of dollars of invested capital, then the service may not be cheap at all.

It may simply be a premium transportation product.

The investment overhang is enormous

Waymo is not just trying to cover the operating cost of each ride. It also sits on top of years of research and development spending. Alphabet has invested heavily in autonomous driving for well over a decade.

That historical investment may be treated differently from an accounting perspective, but economically it still matters. Investors do not fund projects forever out of charity. Eventually, the business must either generate meaningful profit, strategic value, or market power.

This creates a very high bar.

A small fleet of profitable vehicles would not be enough. Even if Waymo eventually makes money on each ride in Phoenix, San Francisco, Los Angeles, Austin, Atlanta, or other markets, that does not automatically justify the total investment.

To make a real dent in billions of sunk and ongoing costs, Waymo would need enormous scale: likely tens of thousands of vehicles just to become a serious transportation business, and hundreds of thousands or more to become the kind of platform that justifies the original vision.

But scaling is not free. Each city adds complexity. Each vehicle adds capital cost. Each market brings local regulation, local operations, local maintenance, local support, and local edge cases.

This is not software scaling like a search engine or app. It is software plus hardware plus roads plus weather plus people plus liability.

That is a much harder business.

The better question is not “Will Waymo work?” It is “For whom?”

Waymo may work technically. It may even work economically in certain dense, wealthy, high-demand markets. It may become a useful premium service in cities where parking is painful, Uber is expensive, and consumers are willing to pay extra for a quiet driverless experience.

But that is not the same as transforming transportation for everyone.

The real question is not whether autonomous vehicles can exist.

They can.

The real question is whether they will make ordinary people’s lives cheaper, safer, and easier than the alternatives.

If Waymo is more expensive than Uber, more expensive than Lyft, and dramatically more expensive than the marginal cost of using a personally owned vehicle, then it is hard to call it democratization.

If the safety gains can be achieved more broadly through mandatory vehicle safety technology, better road design, better enforcement, and better regulation of human-driven fleets, then safety alone does not justify the entire robotaxi project.

If the business only works by charging riders enough to recover massive capital expenditures and support a complex AI-and-operations machine, then the result may be less like public mobility and more like privatized metered access to movement.

And if that is the outcome, then people are right to be skeptical.

Because the point of transportation policy should not be to create a world where every trip becomes a revenue event for a tech company.

The point should be to help people get where they need to go safely, affordably, and with dignity.

If autonomous vehicles can do that better than the alternatives, they deserve a place.

But if they cannot, then the question becomes painfully simple:

All of this, for what?