Days after taking my first Waymo trip in San Francisco in 2024, I wrote about the missed opportunity hiding in plain sight: the empty robotaxis endlessly circling the city’s streets, and what urban mobility could look like if we put them to more use. Two years later, even as robotaxi fleets expanded and their ridership has skyrocketed, not much has changed. A recent analysis of Waymo’s operations in California demonstrates that while the robotaxis’ “deadheading” – the in-service miles driven without a fare-paying passenger onboard – has decreased as service expanded, it remains far from optimal. Waymo robotaxis in commercial service had a passenger on board only 54% of the time and spent an average of 18 minutes between consecutive passenger trips completely vacant. On top of this chronic underutilization, all major autonomous ride-hailing operators in the United States, including Waymo, operate fleets of fully Electric Vehicles (EVs). So I found myself wondering: as the market share of driverless EVs that neither park nor contribute to gas tax revenue continues to grow, who foots the bill for our public roadway infrastructure? 

Identifying a structurally stable and politically acceptable substitute for the gas tax as the transition to EVs continues has been a hot topic in transportation policy circles in recent years. The House Transportation and Infrastructure committee’s multi-year surface transportation reauthorization bill, a rare bipartisan major bill within the chamber, includes an annual registration fee on EVs. While this issue is not specific to robotaxis, the confluence of this policy gap and the rampant propensity for empty cruising exhibited by current robotaxi deployments creates a situation in which a shrinking segment of the public will subsidize the operations of large corporations.  

Take Waymo’s fleet in California as an example. As of August 2025, public records show the company had 1,429 vehicles in the State. While there is no clear information on how many of those vehicles were deployed in commercial service at a given time throughout the year, the data shows that Waymo’s commercial operations in 2025 have resulted in over 66.7 million vehicle miles traveled, or 46,675 miles per vehicle if we assume the entire fleet was in commercial service. This mileage is roughly four times what the average California driver logs in a year. On top of that, Waymo vehicles have spent a total of 3.5 million hours waiting between one drop-off and the next pick-up in busy California downtowns, and based on publicly available information, none of that waiting occurred in designated parking spaces. 

A similar number of Californians driving non-electric SUVs with a fuel economy of about 22 miles per gallon would have contributed over $2.7 million via the federal excise tax and state sales tax on gas or equivalent roadway usage. Likewise, parking a vehicle for the same number of hours of empty cruising between trips, based on prevailing rates in downtown San Francisco and Los Angeles, where the majority of Waymo’s California operations took place, would have generated anywhere from $17 million to $27 million. Keep in mind that these are potential uncaptured fees for a robotaxi fleet operated by a single operator that constitutes less than 0.005% of all registered automobiles in California, so this issue will cascade rapidly as surging robotaxi fleets replace a larger share of human-driven vehicles. 

None of this should be construed as a categorical objection to autonomous vehicles (AVs). The industry leader in the United States, Waymo, offers a premium product powered by an impressive suite of technologies, and their reporting demonstrates an admirable safety record. The issue is not whether robotaxis should exist; it is that the pricing model we use to fund streets, curbs, and pavements was built around gas-guzzling vehicles that must also be parked somewhere when not in use. Current-day robotaxis – and possibly future personally-owned AVs – do neither. Without a thoughtful legislative response at the state, local, and federal levels, we are heading towards a future where the vehicles that use our busiest streets the most will also be the ones contributing the least towards maintaining them. 

The fix need not be exotic. A mileage-based road use fee for commercial AV fleets, along the lines of road usage charge pilots in California and Oregon, would effectively and proportionally replace the gas tax based on actual usage while disincentivizing empty cruising between robotaxi passenger trips. A flat fee on all EVs, as in the current House proposal, would not. On the contrary, it would treat a low-mileage household EV the same as vehicles in commercial fleets. Commercial robotaxi fleets are an ideal first case for a mileage-based road use fee: the vehicles are centrally managed, their mileage is already tracked and reported, and their business depends on intensive use of public roadways.  

We have spent the past few decades pricing fuel and parking because we decided, collectively, that maintaining roadways and curb space in a state of good repair is a shared responsibility. The vehicles occupying the public right-of-way – and the technology behind them – have outpaced this pricing model. It is time to update the meter. 

Awad Abdelhalim is the Associate Director of the MIT Urban Mobility Lab and Transit Lab. His research leverages urban computing and artificial intelligence to optimize transit operations, deploy emerging technologies, and promote the public interest.  

Disclosure: Generative artificial intelligence from Grammarly and Claude was used to improve the structure and grammar of the presented text. No content was AI-generated. The author does not have any conflicts to declare.