Here's a number that should end every argument about autonomous driving: 450,000. That's how many paid rides Waymo completes every week, as of December 2025. Not test rides. Not demo rides. Paid rides, from real passengers, going to real destinations, with no human driver in the car. In seven US cities: Phoenix, San Francisco, Los Angeles, Atlanta, Austin, Miami, and the broader Bay Area. By February 2026, Waymo had logged 200 million fully autonomous miles on public roads. They'd just raised a $16 billion funding round that valued the company at $126 billion. And they're accelerating. Meanwhile, Tesla's "Full Self-Driving" remains Level 2 — meaning a human driver must supervise it at all times. Cruise, once Waymo's most credible competitor, was effectively killed by GM in December 2024. And the dozen-plus other autonomous driving startups that raised billions over the past decade? Most are dead, pivoted, or running on fumes. The autonomous driving race isn't "heating up." It's over. Waymo won. Here's how it happened and what it means. ## The Long Game Waymo's origin story goes back further than most people realize. It began as the Google Self-Driving Car Project in January 2009, run out of Google X by Sebastian Thrun and Anthony Levandowski. The team had its roots in the Stanford Racing Team that competed in the 2005 DARPA Grand Challenge. That's 17 years of continuous development. Seventeen years of sensor development, mapping, simulation, edge case collection, and real-world testing. When people ask "why is Waymo winning?" the simplest answer is: they started first and never stopped. Google spent $1.1 billion on the project between 2009 and 2015 alone — before it even had a commercial name. The project was spun out as Waymo in December 2016. Since then, it's raised $5.5 billion by 2022, another $5.6 billion in 2024, and $16 billion in February 2026. Total investment: somewhere north of $30 billion over 17 years. That's not venture capital money. That's nation-state money. And it bought something nobody else has: a dataset of real-world driving scenarios that's orders of magnitude larger than any competitor's. ## What 450,000 Rides Per Week Looks Like Let me put the 450,000 rides per week number in context. In October 2024, Waymo was doing about 150,000 paid rides per week across Phoenix, San Francisco, Los Angeles, and Austin. By December 2025 — fourteen months later — that number had tripled to 450,000, having expanded to Atlanta and Miami. For comparison, Uber completed about 28 million trips per day globally in Q4 2025, or roughly 196 million per week. Waymo at 450,000 per week is about 0.2% of Uber's volume. Tiny, right? Except Waymo operates in seven cities. With a fleet of a few thousand vehicles. Uber operates in 10,000+ cities with millions of drivers. Per-vehicle, Waymo's utilization is impressive. And the growth rate — 3x in 14 months — suggests we're still on the steep part of the S-curve. Waymo's rider satisfaction is high. The app has a 4.9-star rating. Users consistently say the rides are smoother than human-driven ones (the car doesn't brake suddenly, doesn't weave, doesn't speed). The main complaints are about pickup logistics — the car sometimes has trouble navigating complex pickup locations like airports or construction zones. ## The Cruise Implosion To understand how dominant Waymo has become, you need to understand what happened to its most serious competitor. Cruise, General Motors' autonomous driving subsidiary, was the number-two player through 2023. They'd raised billions, had a fleet of autonomous vehicles in San Francisco, and were operating a real — if small — robotaxi service. Then, on October 2, 2023, a Cruise vehicle struck a pedestrian in San Francisco who had already been hit by a human-driven vehicle. The Cruise car then dragged the pedestrian about 20 feet. The incident was horrifying. Worse, Cruise initially showed regulators an edited video that omitted the dragging portion. When the full video emerged, the California DMV suspended Cruise's operating permit. Kyle Vogt, the co-founder and CEO, resigned in November 2023. The company suspended all operations. They started returning vehicles to roads in limited testing in May 2024, but the damage was done — not just to their reputation, but to their relationship with GM. In December 2024, GM pulled the plug. CEO Mary Barra announced that GM would stop funding Cruise as an autonomous taxi company and instead fold the technology into GM's existing advanced driver-assistance systems for personal vehicles. After roughly $10 billion in investment, Cruise was essentially dead as a robotaxi company. The lesson: in autonomous driving, trust is everything. One bad incident, compounded by a cover-up attempt, destroyed years of work and billions in investment. Waymo hasn't been immune to incidents — in January 2026, NHTSA and the NTSB opened investigations into Waymo robotaxis for passing stopped school buses and one incident involving a child in a school zone. But Waymo's response has been transparent, and their overall safety record is strong. ## Tesla's Eternal Promise Then there's Tesla. Elon Musk has been promising full self-driving capability since 2016. In October 2016, Tesla's website stated that all Tesla vehicles "have the hardware needed for full self-driving capability at a safety level substantially greater than that of a human driver." Customers paid thousands of dollars for a "Full Self-Driving" option based on this promise. It's now February 2026. Tesla's FSD remains a Level 2 system. The driver must supervise at all times. There's no hands-off, eyes-off driving. There's certainly no driverless operation. Tesla's approach is fundamentally different from Waymo's. Where Waymo uses lidar, radar, and cameras in a multi-sensor fusion approach, Tesla has bet exclusively on cameras and AI — what Musk calls "vision-only." The argument is that humans drive with vision alone, so cameras should be sufficient. The problem is that humans have depth perception, decades of real-world experience, and an intuitive physics engine refined by millions of years of evolution. Cameras have pixels. The vision-only approach has improved dramatically — Tesla FSD version 12 and beyond, powered by end-to-end neural networks, handles many driving scenarios impressively. But it still fails in edge cases: unusual construction zones, ambiguous traffic signals, adverse weather, complex intersections. Tesla has talked about launching a robotaxi service for years. At the "We, Robot" event in October 2024, Musk unveiled the Cybercab concept. But there's no timeline for actual unsupervised commercial service. Tesla would need regulatory approval in each jurisdiction, and no regulator is going to approve a system that still requires human supervision. The contrast is stark. Waymo has 450,000 paid rides per week with no driver in the car. Tesla has an impressive driver-assistance system that requires constant driver attention and no commercial robotaxi service. ## The Technology Gap What separates Waymo from everyone else? It's not one thing. It's the stack. **Sensors.** Waymo uses a proprietary sensor suite that includes lidar, radar, and cameras — all built in-house. Their fifth-generation sensor system, introduced with the Jaguar I-PACE fleet and continued with the Zeekr-based vehicles, provides 360-degree perception with redundancy. If one sensor fails, the others compensate. The in-house lidar, which replaced an early $75,000 Velodyne unit, costs roughly 90% less and provides better performance. **Maps.** Waymo operates on extremely detailed 3D maps of its service areas. These maps include not just road layout but lane markings, curb heights, traffic signal positions, and permanent obstacles. The car knows what the world should look like and can detect when something is different — a construction zone, a fallen tree, a new traffic pattern. **Simulation.** Waymo runs millions of simulated miles per day. Every real-world encounter — a jaywalking pedestrian, a car running a red light, an ambiguous hand gesture from a construction worker — gets replayed in simulation with variations. What if the pedestrian was moving faster? What if there were two pedestrians? What if it was raining? This creates a training dataset of edge cases that would take decades to encounter organically. **200 million real-world miles.** There's no substitute for real-world driving data. Every mile teaches the system something that simulation can't perfectly capture — the way a truck's air brakes hiss at a stoplight, the way a cyclist looks over their shoulder before turning, the way fog forms in a specific intersection at dawn. Waymo has 200 million miles of this. Nobody else is close. ## The Business Model Waymo's current business model is straightforward: you hail a ride through the Waymo One app, a driverless car picks you up, and you pay a fare comparable to a standard rideshare. Prices in Phoenix are generally 10-20% cheaper than Uber; in San Francisco, they're roughly comparable. The economics get interesting at scale. A human Uber driver works 6-8 hours a day, needs breaks, and keeps 70-80% of the fare. A Waymo vehicle can operate 20+ hours a day (with breaks for charging and cleaning) and Waymo keeps 100% of the fare minus operating costs. The operating costs are significant: electricity, cleaning, maintenance, insurance, remote monitoring, and fleet management. And the upfront vehicle cost — each Waymo-equipped vehicle costs significantly more than a standard car. But the per-ride economics improve dramatically with utilization. A vehicle doing 30+ rides per day has very different unit economics than one doing 10. At $126 billion valuation, the market is pricing Waymo as a company that will eventually scale to millions of rides per day across dozens of cities. That's a big bet. But the trajectory — from 100K rides per week in early 2024 to 450K by end of 2025 — suggests the scaling playbook is working. ## What Comes Next Waymo's expansion roadmap includes Tokyo as its first international market, announced in partnership with Nihon Kotsu, Japan's largest taxi company. The Tokyo launch would be the first autonomous robotaxi service outside the US by a Western company. Domestically, the next wave of cities likely includes Washington DC, Seattle, and Chicago. Each new city requires months of mapping, testing, and regulatory approval. Waymo can't just flip a switch. But the process is getting faster as they standardize their launch playbook. The regulatory environment is evolving. Some states have embraced autonomous vehicles; others are cautious. The NHTSA investigations opened in January 2026 over school bus incidents are a reminder that regulatory acceptance isn't permanent — it must be continuously earned through safe operations. ## Why This Matters The autonomous driving industry spent over $100 billion in venture capital, corporate investment, and R&D over the past decade. Dozens of companies competed. Most failed. The survivors can be counted on one hand, and Waymo towers above them all. This matters beyond the transportation industry because it's one of the clearest examples of how AI deployment actually works in the real world. Not overnight disruption. Not a single breakthrough. Seventeen years of grinding, iterating, failing, and improving. Billions of dollars. Thousands of engineers. Millions of test miles. And finally, a product that works well enough that real people pay real money to use it every day. The autonomous driving race taught the tech industry something important: the gap between a demo and a product is measured in years and billions of dollars. Any company can build an autonomous vehicle that works on a closed course. Getting one that works reliably in rain, at night, with construction zones and jaywalking pedestrians and double-parked delivery trucks — that's the hard part. And right now, only one company has figured it out at scale. Waymo. Winner by a mile. Actually, 200 million of them.