When we talk about a technology like autonomous driving — developing at great speed but unevenly, limited to a few markets, and once deemed impossible by many — it’s worth occasionally pausing to take stock of how far we’ve come.
In just 18 months, Waymo has gone from 50,000 to 250,000 paid rides per week across its four US markets (Phoenix, San Francisco Bay Area, Los Angeles, and Austin) — a milestone that confirms the operational maturity of autonomous driving technology and the growing interest of users in a service that is becoming almost routine in some cities.
This growth coincides with a new, still pre-published study analyzing 91 million driverless kilometers, which concludes that the Waymo Driver dramatically reduces serious accidents compared to human driving.
Safety is an unarguable point — because the data is clear: 92% fewer injured pedestrians, 82% fewer injuries to cyclists and motorcyclists, and 96% fewer intersection crashes compared to equivalent human driving averages. Given this stark contrast, it becomes difficult to argue against autonomous driving becoming the default option in dense urban areas sooner or later. The reduction in the social cost of traffic accidents is simply too valuable to ignore.
Those who claimed autonomous driving was unworkable “unless all vehicles are autonomous” were entirely wrong: every autonomous vehicle added to the system improves its safety.
The same study does acknowledge a very specific limitation — likely one of the main global adoption hurdles: each deployment requires high-definition maps that must be created and, more importantly, constantly updated with every possible road event. Academic literature points out that traditional methods are very costly, requiring specialized mapping vehicles and highly trained personnel, and that maps become outdated quickly as cityscapes change or routine maintenance is performed — say, sweeping a street or fixing a pothole.
These hidden costs explain why we’re still talking about four or five cities, not forty or fifty.
Waymo thoroughly maps every new region before launching a taxi service. This dependence on centimeter-accurate maps is the antithesis of the…
Meanwhile, driving is no longer what it once was. It’s important to zoom in on the fact that driving is becoming increasingly unbearable. Speed cameras, lower speed limits, low-emission zones, automated controls and constant surveillance have killed the pleasure of driving for most people. Driving is no longer the synonym for freedom it was once sold as — it’s now a continuous exercise in regulatory compliance. In that context, handing over the wheel to an autonomous system is not just safer: it’s more sensible, more efficient, and likely more enjoyable.
The US government has just relaxed several federal regulations so that manufacturers can request safety exemptions for vehicles that are fully autonomous— a move aimed at “winning the race against China.” The message is clear: Washington is willing to lower barriers if it attracts capital and talent. In the European Union, by contrast, regulatory frameworks remain fragmented and problematic, creating a far less fertile ground for large-scale experimentation.
As the United States fine-tunes its regulations, Chinese company Pony.ai has cut its system costs by 70% and is aiming for profitability by 2025, backed by partnerships with Toyota, BAIC, and GAC, and supported by China’s more homogeneous regulatory environment. This blend of cheap manufacturing and government support could make Chinese operators the de facto suppliers of global autonomous driving, particularly in emerging markets where Waymo or Cruise have no presence.
Waymo isn’t the only US company taking on the Chinese: Uber has struck a deal with May Mobility to deploy thousands of robotaxis, while Aurora’s trucks are already making driverless deliveries between Dallas and Houston. But they all share the same Achilles’ heel: geographically concentrated growth, reliant on expensive mapping and surgical agreements with each local authority.
In Alphabet’s latest earnings call, Sundar Pichai hinted that Waymo is considering selling autonomous vehicles to individuals in the long term. The idea raises concerns: if we shift from a shared fleet model to privately owned autonomous cars, congestion could worsen instead of improving, since the marginal cost of moving a driverless vehicle is very low — and the temptation for “empty trips” (vehicles circulating without passengers) would rise, along with repurposing vehicles for pickups and other uses. It would be a repeat — on steroids — of the phenomenon we already saw during Uber and Lyft’s early expansion.
If anything has become clear from the journey between Phoenix and Austin, it’s that autonomous driving has stopped being an experiment and is increasingly becoming infrastructure. Even so, its global rollout will remain uneven: we’ll see islands of autonomy — well-mapped city centers, logistics corridors — surrounded by areas where human drivers are still essential.
That said, for autonomy to deliver its full social value (fewer crashes, fewer emissions, fewer parked cars, etc.), two conditions must be met: first, keep it within shared fleet models, integrated with public transport and taxed for empty trips; and second, develop open standards for dynamic mapping and data exchange, allowing any operator to update their maps without duplicating every investment.
And if we don’t do it for efficiency or sustainability, let’s at least do it out of boredom: if driving is no longer fun, then let it at least be unnecessary. Because what’s clear is that autonomy has already won the technical and ethical debate — now we just need to embrace it as a society and roll it out wisely.
* Enrique Dans is Professor of Innovation at IE Business School and blogger (in English here and in Spanish at enriquedans.com)
Source: Medium
