Bussiness
Waymo and Tesla have opposite problems as they compete for driverless tech dominance
Two major players — Tesla and Waymo — are battling for dominance in the driverless tech sector. At stake is a controlling share of a growing multibillion-dollar market.
But each company has opposite hurdles to clear before a winner can be determined.
With roughly 700 autonomous cars already accepting passengers in San Francisco, Los Angeles, Phoenix, and Austin, Waymo has an early advantage. The company has an established track record of its driverless technology software working well in congested city traffic.
Founded in 2009 as a subsidiary of Google’s parent company, Alphabet — giving it the resources of a multitrillion-dollar company — some analysts argue Waymo is “playing chess” while Tesla plays checkers.
Others aren’t so sure.
Andrej Karpathy, a founding team member of OpenAI and former senior director of AI at Tesla, said in a recent episode of the “No Priors: Artificial Intelligence” podcast that, despite the common refrain that Waymo is ahead of Tesla, he believes Tesla has the advantage.
“I know it doesn’t look like that, but I’m still very bullish on Tesla and its self-driving program,” Karpathy said. “I think that Tesla has a software problem, and I think Waymo has a hardware problem, is the way I put it — and I think software problems are much easier.”
Tesla has succeeded in the widespread deployment of all its cars around the globe, at a scale Waymo can only dream of achieving. So when Tesla works out its autopilot software problems to the point where the tech can actually be deployed, Karpathy said, “It’s going to be really incredible.”
“And so I think Waymo looks like it’s winning right now,” Karpathy said. “But I think when we look in 10 years at who’s actually at scale and where most of the revenue is coming from, I still think Tesla’s ahead in that sense.”
Karpathy also addressed one of the biggest criticisms about Tesla’s autonomous driving system — that it relies on cameras rather than lidar sensors to navigate its environment.
“I’m not sure people are appreciating that Tesla actually does use a lot of expensive sensors,” he said, adding that the EV company uses sensors at a different stage in the deployment process from Waymo. In other words, Tesla uses lidar during the data-gathering process, helping to inform the vehicle’s camera system, which Karpathy said is a more scaleable method.
The miles ahead
Some industry experts say Tesla and Waymo are difficult to compare but for different reasons.
Dan O’Dowd, CEO of The Dawn Project, an advocacy group campaigning against Tesla’s driver assistance software due to consumer safety concerns, argued that Waymo is miles ahead of Elon Musk‘s company.
He pointed to Waymo’s recent deployment of robotaxis to the public in San Francisco and other major cities, where the company has collectively seen 100,000 automated rides a week. Contrast that to Tesla, which has yet to prove its Full-Self Driving system is fully autonomous despite Musk’s repeated proclamations that the company is almost there.
“Tesla has nothing approaching a self-driving car,” he told Business Insider.
Kevin Chen, a former software engineer for the autonomous driving industry who has used both Tesla’s FSD and Waymo, told Business Insider he sees some validity to Karpathy’s sentiment.
He viewed Tesla as a “pioneer” in using machine learning for its autonomous driving software. Given that the company has millions of cars on the road today, he said it’s ahead in manufacturing autonomous hardware at scale.
But he also argued that Tesla’s autonomous hardware hasn’t solved the driverless equation.
“We don’t know if the camera-only approach is sufficient to reach driverless quality with the current state of machine learning and AI,” he said. “So then the big question is: Are you waiting for some AI breakthrough to solve this problem? Or maybe you’re waiting for some chipmaking breakthrough?”
Waymo, on the other hand, produces an autonomous driving system capable of eliminating a human driver. However, the company still relies on remote operators in trickier driving situations that require human intervention.
The engineer added that the two companies are apples and oranges but agrees that Tesla and Waymo face very different problems: Tesla’s autonomous driving system doesn’t work yet, while Waymo’s system works but is prohibitively expensive.
“The sensors and computers on the car — very expensive,” Chen said of Waymo. “Maintaining the map is quite expensive, and in the longer term, they’re also going to have a financial engineering problem where they own and operate the entire fleet. So if you want to massively expand it and you need to hold all those cars on your balance sheet, you’ve got to figure out how to finance those.”
Representatives for Tesla and Waymo did not respond to requests for comment.