Self-driving cars are often pitched as a greener future for transportation, but the environmental picture is far more complicated. Research consistently shows that autonomous vehicles could significantly increase total driving, consume substantial energy through onboard computing, pull riders away from public transit, and accelerate urban sprawl. Even if every self-driving car ran on electricity, these combined effects could easily cancel out the efficiency gains.
People Will Drive a Lot More
The single biggest environmental concern with self-driving cars is simple: they’ll put far more miles on the road. When you don’t have to steer, park, or even be awake, driving becomes almost effortless. That convenience changes behavior in dramatic ways. A naturalistic experiment that simulated private autonomous vehicle use (by giving participants personal chauffeurs) found that people increased their travel by roughly 60%. They took trips they otherwise wouldn’t have, drove farther for errands, and sent their vehicles out without them.
Projections vary widely depending on assumptions, but the direction is consistent. A meta-analysis of 26 studies estimated an overall increase in vehicle miles traveled of about 5.9% with widespread deployment. That figure sounds modest, but it reflects an average across cautious scenarios. When researchers model full automation without policy guardrails, the numbers climb steeply. Studies project increases of 30% to 60% for privately owned self-driving cars in California by 2050, and one Chinese study projected up to a 47% increase in total driving. The European Environment Agency’s 2022 report on shared autonomous vehicles found a median projected increase of 43%.
A major contributor to these extra miles is “empty cruising,” vehicles driving around with nobody inside. Self-driving taxis need to reposition between pickups, and privately owned cars might circle the block rather than pay for parking. Simulations of city-scale ride-hailing fleets estimate that 14% to 25% of all miles driven are completely empty. That’s vehicle exhaust, tire wear, and energy consumption producing zero transportation value.
Onboard Computers Eat Through Energy
A self-driving car isn’t just a car. It’s a mobile data center. The sensors, cameras, and processors needed to navigate roads autonomously consume significant power. Researchers at MIT estimated that the onboard computer in a typical autonomous vehicle draws about 840 watts, roughly the same as running a small window air conditioner continuously while the car moves.
That might not sound like much for a single car, but scaled up, it’s enormous. The MIT team calculated that one billion autonomous vehicles, each driving just one hour per day, would generate emissions comparable to all the world’s data centers combined. Data centers currently account for about 0.3% of global greenhouse gas emissions, roughly equal to the entire country of Argentina. A self-driving car with 10 cameras running 10 neural networks makes about 21.6 million computational decisions every day. Multiply that by a billion vehicles and you reach 21.6 quadrillion inferences daily.
For electric self-driving cars, this computing load directly reduces driving range. Every watt powering the AI is a watt not moving the wheels, meaning more frequent charging and greater total electricity demand. If that electricity comes from fossil fuels, the carbon cost compounds.
Public Transit Takes the Hit
Self-driving ride-hailing services compete directly with buses and trains, and that’s a problem. Public transit moves large numbers of people with far lower per-passenger emissions than any car, even an electric one. When convenient autonomous taxis siphon riders off transit systems, the environmental math gets worse for everyone.
Research from the Bureau of Transportation Statistics found that about 31% of ride-hailing trips directly replace trips that could have been taken on public transit. Roughly 90% of ride-hailing trips had a transit-equivalent option that cost less but took longer. Self-driving taxis will only sharpen that tradeoff. Without the cost of a human driver, autonomous rides will become cheaper, making the speed advantage over transit even more tempting. Every rider who switches from a bus carrying 40 people to a car carrying one dramatically increases per-trip emissions.
Transit systems also depend on ridership to justify service frequency. As riders leave, agencies cut routes, which pushes more people into cars, creating a cycle that’s difficult to reverse.
Sprawl Spreads Further
Self-driving cars change how people think about distance. If your commute becomes productive time where you can work, read, or sleep, a 90-minute drive feels very different than it does behind the wheel. That psychological shift has real consequences for where people choose to live.
Surveys of car-oriented adults found that 29.8% of Americans and 42.3% of Chinese respondents said they’d consider moving farther from their workplace or city center if they had access to a self-driving car. Modeling for the Dallas-Fort Worth metro area predicted that private autonomous vehicle ownership could increase the city’s horizontal spread by 68%. People with longer existing commutes were especially likely to consider relocating even farther out.
Urban sprawl is one of the most carbon-intensive land use patterns. It replaces forests and farmland with roads and rooftops, increases per-capita driving, makes public transit impractical, and locks communities into car dependency for decades. Self-driving technology doesn’t cause sprawl on its own, but it removes one of the strongest natural brakes on it: the unpleasantness of a long drive.
The Weight of Extra Hardware
Every self-driving car carries equipment that a conventional vehicle doesn’t: rotating laser scanners, multiple radar units, an array of cameras, ultrasonic sensors, GPS receivers, and a high-performance computing unit to tie it all together. This hardware adds weight, which increases energy consumption per mile regardless of whether the car runs on gasoline or batteries.
Manufacturing these components also carries an environmental cost. Silicon chips are a well-documented environmental hotspot in electronics production. Life cycle assessments consistently identify chip fabrication as the most resource-intensive stage, requiring ultrapure water, rare chemicals, and enormous amounts of energy. Self-driving cars need far more powerful processors than standard vehicles, and they need several of them for redundancy. As this hardware ages and needs replacement on cycles faster than the car itself wears out, it creates a recurring stream of specialized electronic waste that current recycling systems aren’t designed to handle.
Why Electrification Alone Won’t Fix It
The most common counterargument is that self-driving cars will be electric, so the emissions problem solves itself. But electrification only addresses tailpipe emissions. It does nothing about the 30% to 60% increase in total driving, the energy demands of onboard computing, the loss of transit ridership, or the carbon locked into sprawling development patterns.
More miles driven means more electricity consumed, and in most countries, the grid still relies partly on fossil fuels. More cars on the road, even electric ones, means more tire and brake particulate pollution, more road infrastructure, and more congestion. An electric self-driving car circling empty while its owner shops produces zero local exhaust but still consumes energy, generates particulate matter from tires, and occupies road space that could serve higher-capacity transportation.
The core environmental risk of self-driving cars isn’t the technology in any single vehicle. It’s the behavioral and systemic changes that follow: more driving, fewer transit riders, longer commutes, and cities that spread further into natural landscapes. Without deliberate policies to manage these effects, the efficiency gains of autonomous driving could be overwhelmed by the sheer increase in how much we drive.