I’ve seen thousands of pitches over the years. As an angel investor in over 200 companies, including some of the foundational players in AI like OpenAI, Anthropic, and Scale AI, my inbox is a constant flood of “the next big thing.” For the last decade, a huge portion of that has been autonomous vehicles. I remember one pitch, around 2016. A team of brilliant, starry-eyed engineers showed me a slick video of a family sleeping in their car as it cruised down a moonlit highway. Their tagline was something like, "Your commute is now your bedroom." It was a beautiful vision. I passed on the investment.
I’ve been promised that world since I was a kid, reading Isaac Asimov. We all have. We were told by major CEOs that by 2020, then 2022, and now with the goalposts ever-shifting, that our cars would be driving us while we sleep. It’s a fantastic dream. It’s also a complete dead end.
Let’s be clear: the pursuit of the Level 4 and 5 autonomous vehicle as a mass-market consumer product is a money pit, a technical nightmare, and a regulatory black hole. The tens of billions of dollars poured into this dream have yielded little more than glorified, and sometimes dangerously unreliable, driver-assist systems. We are not on a steady path to success; we are stuck in a loop of diminishing returns. We’re no closer to true, ubiquitous self-driving than we were five years ago. Why?
The Infinite, Unsolvable Edge Cases
Driving in the real world isn’t a clean data set. It’s not a game of Go or a protein-folding problem. It’s a chaotic, unpredictable, and deeply human mess. A computer can be trained on millions of miles of road data, but it can’t be trained for the infinite strangeness of reality. The kid who runs into the street chasing a ball. The construction worker waving a confusing, non-standard hand signal. A plastic bag that flaps in the wind just enough to look like a deer. A puddle that’s actually a deep, axle-breaking pothole.
These aren’t “edge cases” that can be patched with a software update. They are the fundamental nature of an open environment. Every time a developer thinks they’ve accounted for one, a thousand new ones appear. It’s a game of whack-a-mole where the moles are breeding exponentially.
I’ve sat in the boardrooms and seen the demos. The tech is impressive in a controlled setting. But it’s brittle. It works 99.9% of the time, and everyone celebrates. But that 0.1% is the difference between a safe trip and a fatal, headline-grabbing accident. The liability question is a Gordian knot no one wants to touch. Who’s at fault when a car decides to swerve into oncoming traffic? The owner who wasn’t paying attention? The manufacturer? The software developer who wrote the perception code? The company that supplied the LiDAR sensor? No one has a good answer, because there isn’t one.
We’ve been so mesmerized by the sci-fi vision of robotaxis that we’ve ignored the real AI revolution happening right in front of us. It’s not on our highways. It’s in our factories, our warehouses, and our operating rooms.
Where the Real Robots Are (And Where the Real Money Is)
While everyone was watching Waymo and Cruise burn through cash in the streets of Phoenix and San Francisco, a quieter, more profound transformation was taking place. I’m talking about physical AI in controlled environments. This is where the real value is being created, and it’s where I’m putting my money.
This isn’t a new idea. It’s the logical evolution of automation. The difference is that AI and robotics have finally reached a point of sophistication where they can tackle complex physical tasks, as long as the environment is constrained. The variables are known. The goals are clear. The ROI is measurable in months, not decades.
Think about it:
Warehouse Robots: I’ve been an investor in logistics and fulfillment companies for years. The shift is staggering. For decades, warehouses used simple autonomous guided vehicles (AGVs) that followed magnetic strips on the floor. Now, with advanced AI, companies like Amazon Robotics (which acquired one of my early investments, Kiva Systems) have fleets of thousands of robots that can navigate dynamically, pick, pack, and ship orders with incredible speed and accuracy. The environment is controlled, the tasks are repetitive, and the ROI is massive. There are no pedestrians, rogue plastic bags, or confusing hand signals in a warehouse aisle.
Surgical Robots: Would you trust a robot to perform surgery on you? Millions of people already have. I remember first seeing a demo of the Da Vinci surgical system and being blown away. It’s not a robot operating on its own. It’s a tool that gives a surgeon superpowers. It has been used in millions of procedures, allowing surgeons to operate with a level of precision and control that is humanly impossible. The robot isn’t replacing the surgeon; it’s augmenting their skill in the most controlled, predictable environment imaginable: the human body, mapped and understood before the first incision is made.
Humanoid Robots: This is the next frontier, and it’s the one I’m most excited about. Forget the car. The most useful form factor for a robot is the one our entire world is built for: the human form. Companies like Figure AI, which I’m incredibly excited to be an investor in, are building humanoid robots that can perform labor in places designed for people. They can walk up stairs, open doors, and use human tools without needing to rebuild the entire factory. The initial applications aren’t in our homes, but in manufacturing plants, logistics centers, and disaster sites. They are tackling critical labor shortages and doing the dangerous jobs humans shouldn’t have to.
Even Tesla, for all its grand pronouncements about Full Self-Driving, seems to understand this on some level. Their most interesting project right now isn’t the Cybertruck; it’s Optimus. They know that a general-purpose humanoid robot that can operate in a factory has a much clearer and more immediate path to profitability than a car that can navigate the chaos of downtown San Francisco during rush hour.
Stop Chasing the Wrong Dream
The dream of the autonomous car is seductive. It promises a future of leisure and convenience, a solution to traffic and accidents. But the path to that future is littered with technical and ethical roadblocks that we have no clear path to solving. We’ve become so fixated on this one, incredibly difficult application of AI that we’ve failed to see the incredible progress happening elsewhere.
The real robotics revolution isn’t about replacing the driver in an uncontrolled, unpredictable world. It’s about automating the dull, dirty, and dangerous jobs that humans shouldn’t be doing. It’s about creating tools that make us more productive, more precise, and safer in our work.
So, the next time you hear a founder promising a fleet of robotaxis by next year, ask them the hard questions. Ask them about the long tail of edge cases. Ask them about liability. And then ask them why they aren’t building something useful instead. The future of AI is physical, but it’s not on the open road. It’s in the controlled, structured environments where robots can deliver real, tangible value today. That’s where the next generation of great companies will be built.
Frequently Asked Questions
What experience informs this perspective?
This perspective comes from over a decade of building companies in Silicon Valley, two successful exits (RemoteTeam to Gusto, MovieLaLa to Gfycat), and investing in 200+ startups including Anthropic, OpenAI, and Scale AI. I write about what I've lived.
How can I apply this thinking to my own situation?
Start by identifying the core principle behind the opinion, not the specific example. Then ask yourself: does this principle apply to my context? If yes, test it in a small, low-risk way before going all in.
What's the most common pushback you get on this?
People often push back by citing exceptions or edge cases. And they're usually right that exceptions exist. But building a strategy around exceptions rather than patterns is a losing game for most founders.
Do all experts agree with this view?
No, and that's fine. The best ideas in business are often contrarian. I share my perspective based on my experience and data, but I encourage you to seek out opposing viewpoints and form your own conclusions.