“Fully autonomous vehicles are five years away” has been the prediction for about twenty years now. It’s become a joke in tech circles - autonomy is perpetually near but never here.

Except something has changed. Waymo operates actual commercial robotaxi service in multiple US cities. Cruise stumbled but is recovering. Chinese companies like Baidu’s Apollo and Pony.ai are deploying at scale in China.

So where do we actually stand? Let me try to cut through the hype in both directions.

What’s Actually Working

Geofenced robotaxis are real. Waymo operates paid rides in Phoenix, San Francisco, Los Angeles, and Austin. You can hail a car with no driver and get where you’re going. Hundreds of thousands of rides have been completed.

This isn’t a publicity stunt. It’s a commercial operation with paying customers. The service works in defined areas under defined conditions. That’s genuine progress.

Highway ADAS (Advanced Driver Assistance Systems) has become quite good. Tesla’s Full Self-Driving, GM’s Super Cruise, Ford’s BlueCruise, and similar systems can handle highway driving with decreasing driver intervention required. They’re not autonomous - they require attention and readiness to take over - but they’re useful.

Trucking on defined routes is advancing. Companies like Aurora, Waymo Via, and Kodiak are running autonomous trucks on highway routes. The economic case is compelling (driver shortage, labor costs), and highway driving is more structured than urban environments.

Controlled environments work well. Ports, mines, warehouses - places with defined paths, controlled access, and no unexpected pedestrians - have successful autonomous vehicle deployments.

What’s Not Working (Yet)

True L5 autonomy - a vehicle that can drive anywhere a human can drive with no restrictions - doesn’t exist and isn’t close. The long tail of edge cases in unstructured environments remains unsolved.

Scaling beyond geofences has been harder than expected. Waymo’s service areas, while expanding, are still limited. The operational overhead of mapping and validating new areas is substantial.

Adverse conditions remain challenging. Heavy rain, snow, unusual construction, aggressive human drivers - these create situations where autonomous systems struggle.

Regulatory patchwork creates deployment challenges. Rules vary by state, by country, and sometimes by city. Each jurisdiction requires separate approval processes.

Unit economics at scale are still unproven. Waymo and others are operating, but the cost per ride and path to profitability remain unclear at the vehicle volumes currently deployed.

The Business Reality

The autonomous vehicle market has consolidated significantly. The 2019 landscape had dozens of well-funded startups. Now, a handful of serious players remain:

Waymo has Google’s deep pockets and patience. They’ve invested more time and money in autonomy than anyone. Their conservative, systematic approach has yielded an actual operating service.

Cruise (GM) stumbled badly with a pedestrian incident and had to pause operations. They’re recovering but lost significant momentum and trust.

Aurora is focusing on trucking, where the economics are more favorable and the environments more controlled.

Tesla takes a different approach - vision-only autonomy deployed to millions of vehicles, with continuous learning from real-world driving. More aggressive and controversial, but generating massive amounts of real-world data.

Chinese companies (Baidu Apollo, Pony.ai, WeRide, AutoX) are deploying in China where regulatory support is strong and competition for the domestic market is intense.

Many startups have shut down, been acquired, or pivoted. The capital requirements and timeline to commercialization proved too much.

Investment Implications

For innovation managers and investors:

Component suppliers may be safer bets than full-stack AV companies. LiDAR, radar, compute platforms, simulation software - these sell regardless of which AV companies succeed.

ADAS is the near-term opportunity. Driver assistance systems that don’t claim full autonomy face lower regulatory barriers and clear consumer demand. This market is growing now.

Trucking is closer than robotaxis. The controlled-route, highway-focused use case has better near-term economics and more achievable technical requirements.

Geographic matters. Regulatory environment differs dramatically. US, China, and to some extent Europe are the markets where deployment is happening. Other regions lag significantly.

Watch the safety data. As autonomous vehicles accumulate more miles, we’ll have better data on their actual safety compared to human drivers. This data will influence regulation and adoption.

The Honest Timeline

If I had to bet:

Now - 2027: Continued expansion of geofenced robotaxi service to more cities, with gradual expansion of service areas. Highway autonomous trucking pilots becoming more common.

2027 - 2030: Significantly broader robotaxi availability, potentially in most major US metropolitan areas. Trucking moving toward commercial deployment at scale.

2030+: The question is whether true L5 autonomy ever arrives, or whether we plateau at highly capable L4 systems that work great in defined conditions but never quite reach everywhere/anytime capability.

The technology has clearly advanced. Real services exist. But the grand vision of autonomous vehicles everywhere, replacing all human driving, remains far off. The industry has matured into a realistic assessment of what’s achievable.

For planners in transportation, logistics, and urban planning, autonomous vehicles are coming - but as a gradual transition, not a sudden revolution. Plan accordingly.