Self-Driving Empires: Who's Really Winning the FSD Race?

Tesla sells cars with FSD capability as an $8,000 add-on. You still need to supervise. Hands on the wheel. Eyes on the road. But the car handles most driving tasks.

Their approach is different. Instead of geofencing to specific cities, Tesla wants their system to work anywhere. And instead of building a robotaxi fleet, they're selling the capability to millions of existing customers.

Tesla has something nobody else has—scale. Millions of cars collect data every day. That data trains their neural networks, which theoretically improves the system faster than competitors who only have thousands of vehicles.

Version 14.2 just rolled out. It handles emergency vehicles better, recognizes pedestrian gestures, and navigates construction zones more smoothly. 

Tesla also got a permit for driverless testing in Arizona and is expanding to South Korea, Australia, and New Zealand.

But there's a problem. "Supervised" self-driving isn't really autonomous. And regulators are cautious about approving truly driverless operation for consumer vehicles. Elon Musk keeps teasing that unsupervised FSD is "approaching," but he's been saying that for years.

Waymo runs robotaxis in multiple cities. 

No safety driver. No steering wheel intervention. Just you and an empty front seat.

They're not selling cars to consumers. Instead, they operate like a traditional taxi service, except the taxi drives itself. You open an app, request a ride, and a white Jaguar I-Pace pulls up with nobody inside.

Waymo has something nobody else does—regulatory approval to operate without human oversight. That's huge. Getting that approval takes years of testing and proving safety records to skeptical city officials.

They're opening six new cities this year: Las Vegas, San Diego, Detroit, Minneapolis, Tampa, and New Orleans. They're also building a 239,000 square-foot factory in Phoenix and doubling their fleet to 3,500 vehicles by 2026.

Each Waymo car costs around $200,000. That's $50,000 for the Jaguar, plus another $150,000 in sensors, computers, and self-driving tech. Those economics need to improve before this becomes truly profitable.

Uber doesn't build self-driving cars. They integrate them into their platform. Partner with Waymo? Check. Work with Aurora for freight? Check. 

Now they're building their own fleet with Nvidia technology.

Uber already has the demand side figured out. Millions of people use their app daily. They don't need to convince anyone to try robotaxis—they just need to swap human drivers for autonomous ones.

They're deploying 100,000 Nvidia-powered vehicles by 2026. They're operating in four U.S. cities now, targeting ten by year-end. And in places where Uber already offers Waymo rides, those autonomous trips have higher utilization rates than human drivers.

But their CEO said something important: don't expect self-driving profitability anytime soon. They're balancing massive capital expenditure with growth targets. This is a long-term play.

Aurora focuses on freight—big rigs hauling goods across highways. They're also building a "Driver-as-a-Service" model where companies subscribe to their autonomous system instead of buying it outright.

Long-haul trucking might actually be easier to automate than city driving. Highways are more predictable. Fewer pedestrians. Clearer lane markings. And there's a real shortage of human truck drivers.

Aurora just passed 100,000 driverless miles on public roads. They're expanding their Fort Worth to El Paso route and planning to deploy hundreds of trucks with PACCAR by 2026. Their new hardware costs half as much and lasts longer.

The subscription model is smart. Instead of trucking companies spending hundreds of thousands upfront, they pay monthly. That lowers the barrier to adoption.

Zoox builds purpose-made autonomous cars. No steering wheel. No pedals. These vehicles were designed from scratch for self-driving.

Amazon owns them, which provides deep pockets and logistics expertise. They're targeting urban ride-hailing—short trips in dense cities.

Starting from scratch means they can optimize everything for autonomous operation. Passengers sit facing each other. Doors open from both sides. It looks like a rolling lounge.

But this approach has risks. Purpose-built vehicles are expensive to develop and manufacture. And they need the same regulatory approvals as everyone else, except they can't fall back on having a human driver take over.

Here's the practical breakdown.

Short term (next 2 years): Don't expect profits. All five companies are spending heavily on technology, manufacturing, and regulatory approval. This is the infrastructure-building phase.

Medium term (2026-2028): We'll see commercial scale. Aurora's trucking operations. Waymo in a dozen cities. Tesla potentially unsupervised in some regions. Uber's network effects are kicking in.

Long term (beyond 2028): This is when unit economics matters. Can Waymo get vehicles below $100,000? Does Tesla achieve true autonomy? Can Zoox manufacture at scale?

The industry moved past the pilot phase. 

These aren't tech sci-fi anymore. But investor returns lag operational deployment by years. Maybe many years.

The safety standard keeps rising. Autonomous vehicles don't just need to match human safety—they need to beat it convincingly. 

Otherwise, every accident becomes front-page news and regulatory pushback.

The race is on. But it's a marathon, not a sprint.

Waymo vs. Tesla: