For most of the last decade, self-driving cars were a demo — an impressive video, a limited pilot, a promise perpetually a year or two away. That’s no longer accurate. Waymo now runs roughly 500,000 paid public rides every week across eleven US metro areas, without a driver behind the wheel. Morgan Stanley has called 2026 the industry’s “singularity moment,” projecting that autonomous driving availability in the US will more than double, from 15% of the urban population at the end of 2025 to over 30% by the end of this year.
This isn’t a piece about the far future of driving. It’s about what’s actually operating on roads right now, who’s genuinely ahead, where the real safety debates stand, and what it means for how an ordinary commute might look over the next few years.
Robotaxis Went From Pilot Program to Business Line
The clearest sign that autonomous driving has crossed a real threshold is scale. Waymo alone logs close to 4 million autonomous miles weekly with around 3,000 vehicles, and is targeting 1 million weekly paid rides by the end of this year — up from roughly 400,000 weekly trips earlier in 2026. Its post-money valuation has climbed to $126 billion, which reflects investor confidence that this is now a durable business, not a research project subsidized indefinitely.
Tesla’s approach looks different but is also scaling. After rolling the first fully-autonomous, steering-wheel-free Cybercab off its production line, Tesla is expanding its robotaxi program to seven new cities in the first half of 2026, up from just Austin and San Francisco. Tesla has also crossed ten billion self-driving miles in total. The company’s fleet remains far smaller than Waymo’s — measured in the dozens of vehicles per city rather than thousands — but the underlying technology stack is being validated at genuinely large scale through its consumer Full Self-Driving product.
China is moving just as fast through a different model. Robotaxi fleets there have roughly tripled in a year, and at the 2026 Beijing Auto Show, 23 Chinese cities opened legal Level 3 highway sections with clearly defined accident liability rules — a regulatory clarity that much of the rest of the world is still working toward.
The Shift From L2+ to Genuine Hands-Free Driving
Most consumer vehicles on the road today sit at Level 2 to Level 2+ autonomy — meaning the car manages steering, acceleration, and braking, but the driver must stay attentive and ready to intervene at any moment. The meaningful shift happening in 2026 is the move toward Level 3, where the driver can genuinely disengage — hands and eyes off — under specific, defined conditions.
In China specifically, the narrative has visibly shifted from L2+ refinement to real Level 3 deployment, with OEMs like Chery and suppliers like Bosch pushing hands-free highway driving into production. In the West, the timeline is more staggered: Ford’s Level 3 “eyes-off” driving is targeting a 2028 launch on a new $30,000 platform, with an AI driving assistant arriving sooner, in 2026. Audi’s 2026 A6 and Q6 e-tron models take an incremental approach, layering an augmented-reality system over Level 2 ADAS to overlay navigation and hazard information directly onto the driver’s view.
The Safety Debate Is Real, and the Data Cuts Both Ways
Safety remains the most genuinely contested part of this story, and the data doesn’t point in one clean direction. Waymo claims its driving system is involved in 92% fewer crashes causing serious or fatal injuries than human drivers under comparable conditions, based on over 170 million fully autonomous miles — a substantial and credible sample size. At the same time, notable incidents continue: a Waymo vehicle struck a child at low speed near a school, and other autonomous vehicles have blocked emergency vehicles from passing.
Tesla’s safety picture is similarly mixed depending on which numbers you look at. Its Full Self-Driving (Supervised) system reportedly averages one major collision every 5.3 million miles, which would significantly outperform the human driving average — but separate incident data shows a cluster of lower-speed crashes in recent months, including collisions with a fixed object, a stationary bus, and a heavy truck, along with low-speed backing incidents. Context matters here: NHTSA estimates that roughly 94% of all car crashes involve human error, which is the baseline these systems are actually competing against, not a hypothetical accident-free standard.
The Market Numbers Reflect How Fast This Is Moving
The global autonomous vehicle market is projected to reach roughly $626.9 billion in 2026, continuing a growth trajectory that’s expected to push past $1 trillion by 2028. Within the US specifically, autonomous ride volume is projected to jump from around 15 million rides in 2025 to roughly 36 million in 2026 — a more than doubling in a single year. Longer-term robotaxi forecasts from Goldman Sachs, McKinsey, and BCG converge on a $300 to $400 billion global robotaxi market by 2035, with China alone expected to represent roughly $61.2 billion of that figure.
An important nuance often missed in robotaxi headlines: personal ownership of autonomy-equipped vehicles still accounts for about 74% of total market value today, driven largely by subscription products like Tesla’s Full Self-Driving package priced near $199 a month. Shared robotaxi mobility is the smaller slice of the market currently, but it’s the faster-growing one, forecast to expand at over 25% annually as unit economics improve — meaning the private-ownership and shared-robotaxi models are both scaling, just on different timelines.
What’s Actually Changing About a Normal Commute
- In select major metros, hailing a fully driverless ride is now a routine option, not a novelty — Waymo operates in eleven US cities with high enough volume to function as genuine transportation infrastructure.
- On highways, more mid-range and luxury vehicles now support genuine hands-free driving under specific conditions, shifting the driver’s role from constant control to supervision.
- In China, legally defined Level 3 zones on highways mean some drivers can now legitimately disengage from active driving for portions of a commute, with liability rules already in place.
- Consumer-facing AI driving assistants are becoming a distinct product category, layered on top of existing ADAS rather than replacing it outright.
- Insurance and liability frameworks are still catching up unevenly by region — China’s new L3 liability rules are more defined than most Western jurisdictions currently offer.
What Should You Actually Do With This?
If you’re in a city where Waymo or a similar robotaxi service operates, it’s genuinely worth trying as a transportation option now, not a future curiosity — the ride volume numbers suggest a real, tested service rather than an experimental one. If you’re shopping for a new vehicle and want current-generation hands-free capability, look specifically at what conditions (highway-only, specific weather, mapped roads) any “Level 3” claim actually applies to, since the label alone doesn’t guarantee true eyes-off driving everywhere. And regardless of which system you’re evaluating, treat safety claims from any single company with the same scrutiny — ask what baseline they’re comparing against and over how many real-world miles, since both the strongest and weakest safety claims in this space currently come from companies with a direct financial interest in the story.
Frequently Asked Questions
Is a fully self-driving car actually available to buy in 2026?
Not in the sense of true Level 5 autonomy everywhere. Most consumer vehicles offer Level 2 to Level 2+ systems requiring driver attention, while genuine Level 3 “eyes-off” capability is emerging in specific markets and conditions — notably parts of China and select highway zones — with wider Western rollout, like Ford’s, targeted for 2028.
Are robotaxis actually safer than human drivers?
Company-reported data, like Waymo’s claim of 92% fewer serious-injury crashes over 170 million autonomous miles, suggests meaningful safety improvement in tested conditions. However, notable individual incidents continue to occur, and independent, standardized long-term data across all providers and conditions is still developing, so it’s reasonable to treat this as genuinely improving but not yet definitively settled.
Which company is actually leading in autonomous driving right now?
It depends on the metric. Waymo leads clearly in commercial scale and paid ride volume, with roughly 500,000 weekly rides across eleven US cities. Tesla leads in total autonomous miles driven, having crossed ten billion, largely through its consumer Full Self-Driving subscription base. China leads in regulatory clarity and Level 3 highway deployment specifically.
Why do autonomous vehicles still crash if the technology is improving?
Even well-performing systems operate in unpredictable real-world conditions alongside human drivers, pedestrians, and edge cases that are inherently difficult to fully anticipate. The relevant safety comparison isn’t zero accidents — it’s performance relative to human drivers, who NHTSA estimates are responsible for roughly 94% of all crashes, which is the baseline these systems are actually being measured against.
Will owning a personal car with self-driving features remain more common than robotaxis?
For now, yes — personally owned autonomy-equipped vehicles still represent about 74% of total market value today. However, shared robotaxi services are growing faster on a percentage basis, so the balance between the two models is likely to shift over the next decade even if personal ownership remains the larger segment in the near term.
The Bottom Line
Autonomous driving crossed a real threshold in 2026 — not because the technology became perfect, but because it became operational at genuine scale, with real ride volumes, real revenue, and real (if uneven) regulatory frameworks starting to catch up. Waymo’s weekly ride count, Tesla’s mile count, and China’s Level 3 highway deployment all point the same direction: this is no longer a demo. The open questions now aren’t really about whether autonomous driving works — it’s about how fast trust, regulation, and infrastructure can keep pace with a technology that’s already moving faster than most people expected.
