Andrej Karpathy had spent five years working on Tesla’s self-driving program, and he has an interesting perspective on how Tesla’s approach will compare to Waymo’s in the coming years.
Karpathy, who served as Tesla’s AI Director after co-founding OpenAI, recently shared his nuanced take on the autonomous vehicle race between Waymo and Tesla. While acknowledging Waymo’s impressive current capabilities, he raised important questions about what’s happening behind the scenes and argued that Tesla’s strategy may ultimately prove more scalable—even if it takes longer to fully materialize.
Speaking candidly about Waymo’s operations, Karpathy pointed out an aspect of the technology that often goes unnoticed by passengers. “When you look at these cars and there’s no one driving, I also think it’s a little bit deceiving because there are actually very elaborate teleoperation centers of people actually in the loop with these cars,” he said. “I don’t have the full extent of it, but I think there’s more human in the loop than you might expect.”
He elaborated on this point: “There are people somewhere out there basically beaming in from the sky. I don’t actually know that they’re fully in the loop with the driving. I think some of the times they are. But they’re certainly involved and there are people, and in some sense we haven’t actually removed the person. We’ve moved them to somewhere where we can’t see them.”
Despite his observations, Karpathy was quick to note his appreciation for Waymo’s service. “By the way, I love Waymo and I take it all the time,” he said, before adding a caveat about the current state of autonomous driving. “I just think people are sometimes a little bit too naive about some of the progress, and I still think there’s a huge amount of work.”
Karpathy also noted operational limitations that suggest the technology isn’t as universally capable as it might appear. “For example, Waymo can’t go to all the different parts of the city. My suspicion is it’s parts of city where you don’t get a good signal,” he speculated, acknowledging he was theorizing rather than speaking from direct knowledge of Waymo’s technical stack.
However, when it comes to long-term strategy, Karpathy’s bet is firmly on Tesla’s approach. “I think Tesla took a lot more scalable approach,” he stated. “The team is doing extremely well. I’m on the record for predicting how this thing will go, which is Waymo will have early gains because you can package up so many sensors. But I do think Tesla’s taking the more scalable strategy and it’s gonna look a lot more like that.”
He also pushed back on the narrative that autonomous driving has already “arrived,” emphasizing that true success means deployment at scale. “I didn’t want to talk about self-driving as something that took a decade because it didn’t take. It didn’t take yet, if that makes sense. Because one, the start is at 1980, not 10 years ago. And then two, the end is not here yet. When we’re talking about self-driving, usually in my mind it’s self-driving at scale. People don’t have to get a driver’s license, et cetera.”
Karpathy’s perspective offers an interesting counterpoint to recent comments from Uber CEO Dara Khosrowshahi, who had recently said that he believes LiDAR and radar-equipped vehicles like Waymo’s have a better chance of achieving “superhuman levels of safety” in the near term compared to Tesla’s camera-only approach. Khosrowshahi noted that solid-state LiDAR now costs only $400-500, making the multi-sensor approach increasingly viable.
The debate highlights a fundamental fork in the autonomous vehicle industry: Waymo’s sensor-rich approach offers more immediate results and redundancy but at higher hardware costs and complexity, while Tesla’s vision-only system bets on the power of AI and neural networks to replicate human driving abilities with cheaper, more scalable hardware. Karpathy’s insights suggest that while Waymo may dominate the early innings of autonomous driving, Tesla’s approach—if it successfully solves the AI challenges—could win the longer game by being easier and cheaper to deploy at massive scale. The question remains whether Tesla can close the gap before Waymo’s head start becomes insurmountable, or whether falling LiDAR costs will make the hardware advantage less relevant than the software and operational maturity Waymo is building today.