Waymo vs Cruise: two approaches to the same dream

I rode in a Waymo in Phoenix last month. Two weeks later, I rode in a Cruise in San Francisco. Same category of experience, completely different feeling.

Both are robotaxis. Both drive city streets without a human behind the wheel. Both will take you where you want to go while you sit in the back seat trying to decide whether to watch the road or trust the machine.

The approaches are philosophically different, and riding in both back-to-back made the differences physical.

Waymo: the careful grandparent

Waymo’s approach is mapping. Before their cars drive a street, they’ve already mapped it in extreme detail using LIDAR. Every curb. Every lane marking. Every stop sign. The car knows, before it arrives, what the road should look like. Then it uses real-time sensors to compare what it sees to what it expects.

Riding in a Waymo feels like riding with a very careful grandparent. The kind who drives 2 mph below the speed limit. Who comes to a full, complete stop at every stop sign, counting to three before proceeding. Who brakes early, accelerates gently, and takes turns like the car might shatter if pushed too hard.

I felt safe. Deeply safe. The car was cautious in a way that telegraphed competence. Every decision was conservative, deliberate, and correct.

The Waymo also knew things I didn’t. It slowed down approaching an intersection where a pedestrian was about to step off the curb, before the pedestrian was visible to me. The sensors saw around the corner. The pre-mapped environment told the car that pedestrians frequently cross there. The caution was specific and informed.

But it was also slow. A trip that would take a human driver 12 minutes took the Waymo 18. Every right turn involved a full stop, a scan, and a measured acceleration. The car never took a gap that a human would have taken. It waited for the gap to get bigger.

Cruise: the confident teenager

Cruise’s approach is more camera-heavy, more neural-network-driven, more real-time. GM backs them. Their cars (modified Chevy Bolts) operate in San Francisco, a city that’s arguably harder to drive in than Phoenix. Hills. Narrow streets. Double-parked delivery trucks. Cyclists who treat red lights as suggestions.

Riding in a Cruise felt like riding with a confident teenager. The kind who knows they’re good at driving and is maybe a little too comfortable with tight spaces. The car merged smoothly, took turns with conviction, and navigated double-parked situations with a decisiveness that surprised me.

It was faster than the Waymo. More assertive. When a gap appeared in traffic, the Cruise took it. When a lane was blocked, it adjusted quickly. The car felt like it was improvising, responding to the city in real time rather than checking a map.

But there were moments. A hesitation at an intersection that lasted just long enough for me to wonder if the car was confused. A lane change that started, paused, and then committed in a way that felt like the neural network was deciding mid-maneuver. A moment where a cyclist appeared from behind a truck and the car braked harder than seemed necessary.

Different philosophies

Waymo’s philosophy is: know everything before you drive. Map the world. Pre-compute the safe behavior. Then execute it with sensor verification.

Cruise’s philosophy is: learn from what you see. Process visual data in real time. React, adapt, respond. Map is secondary to perception.

Both work. Both are getting people from point A to point B without a human driver. But the experience of being a passenger is different in a way that reveals the underlying architecture.

Waymo’s conservatism is its map showing. The car knows the intersection so well that it can afford to be cautious. It’s not unsure. It’s careful by design.

Cruise’s confidence is its neural network showing. The car is processing what it sees right now and making decisions in real time. It’s not reckless. It’s responsive by design.

Which do I trust more?

Honestly? I don’t know.

I trusted the Waymo in the moment. The caution felt like competence. I never felt unsafe. I never wondered if the car knew what it was doing.

I was more impressed by the Cruise. The assertiveness felt like skill. The car handled situations that I would have found stressful as a driver. It navigated San Francisco better than I would have.

But trust and impression aren’t the same thing. I trusted the careful grandparent. I was impressed by the confident teenager. I’d put my kids in the Waymo. I’d put myself in the Cruise when I was late.

I think this is how self-driving plays out. Not one winner. Not one approach. Different systems for different cities, different contexts, different riders. The future of driving isn’t a single company. It’s a fleet of different answers. Same dream, different paths.

Both paths are closer than most people realize.

Related thinking: