Scalability without stability is an illusion. Waymo is going to Singapore, with commercial service planned for 2028. The car in the photo is a typical multi-sensor vehicle packed with LiDAR, radar, and cameras. Many people look at this and say, “Multi-sensor systems are confident, but Tesla has scalability.” I believe this dichotomy is wrong. Scalability is not about adding more maps. It’s about being able to increase the number of vehicles, operational density, and operating hours while maintaining the same level of safety. Without stability, regulatory hurdles, insurance costs, public opinion, and remote intervention expenses all rise. And then, scalability stops. Let’s start with the numbers. Waymo is operating fully driverless commercial services in 14 metropolitan areas across the U.S., with over 200 million cumulative driverless miles. It has published data showing its accident rate per mile within its operational areas is lower than that of human drivers. That’s why it qualifies to enter new countries like Singapore. Yes, it takes time to rebuild maps and learn monsoons and local road conditions—that’s a drawback. But that time is spent not because it “can’t go,” but because it’s going to ensure it can sustain operations. Tesla’s RoboTaxi is different. Fully driverless operations are currently limited to six cities in Texas and Florida, with the company reporting 1 million cumulative driverless miles. That number jumped rapidly from 380,000 in July to 1 million by early September. The problem? That 1 million miles is only about 0.5% of Waymo’s total. The sample size is far too small to validate failures that occur once every ten million miles. “No notable incidents” is an assertion—not a third-party verified metric based on accident rates per mile. Videos of vehicles passing right by bollards continue to surface. There is data showing Tesla’s supervised FSD for consumer vehicles reduces collisions by 40% compared to manual driving in Australia and New Zealand. But that’s Level 2 driving—with a human driver present. It is not evidence comparable to RoboTaxi scalability. So the conclusion is simple: Multi-sensor systems are expensive and slow. But because they’ve already proven stability in limited areas, they can expand to new cities and countries. Tesla claims its cars are cheap and only software needs to change. The potential is there. But unless driverless operation stability is proven at scale, announcements of new city launches are not scalability—they’re just lists. Cheap hardware is a necessary condition for scalability. It is not a sufficient condition. Stability comes first. Scalability without stability is an illusion.