- Waymo Via is the trucking arm of Waymo, the Alphabet self-driving startup.
- Insider spoke with three of Waymo Via’s leaders about the company.
- Trucking is an $800 billion business in the US, so the opportunity to automate it is huge.
- See more stories on Insider’s business page.
Waymo, formerly known as the Google self-driving-car project, has been known for many things over the past 10 years: Toyota Priuses that can drive themselves, an adorable autonomous pod dubbed the “Firefly,” and, more recently, a ride-hailing service in Arizona with minivans that have no human hands on the wheel.
But somewhat overlooked at the firm, which has been an independent division of Alphabet since 2016, is the company’s trucking arm, Waymo Via.
That’s set to change as Via presses toward commercialization, following the rollout of the Waymo One robotaxi service in 2018. Insider spoke with three of Via’s top executives to get the lowdown on how the firm plans to safely send 80,000-pound fully loaded semis rolling down US highways with no truckers in the cabs.
“The invisible world of trucking and logistics is a huge opportunity for innovation,” Charlie Jatt, Waymo’s head of commercialization for trucking, said.
Jatt is a San Francisco native who graduated from Stanford in engineering physics and then undertook a stint in management consulting at Bain. After General Motors bought then-tiny self-driving startup Cruise in 2016, he decided to explore the gestational self-driving industry and joined Waymo. Two years ago, when the company began to adopt what he calls a “portfolio view” of its business, he moved over the trucking side.
Waymo intends to develop three core autonomous offerings, each constructed around what former CEO John Krafcik termed a “driver” — a suite of hardware and software technologies that functions like a human operator and that could, in theory, drive anything (or fly or sail anything). Think of the Waymo driver as a combination of advanced laser radars and sensors built into vehicles, plus the software that controls them along with the human and technological infrastructure that works with consumers and governments to bring services to market.
Those services, if Waymo can successfully execute, should be: ride-hailing; freight, logistics, and delivery; and “last mile” solutions for public transit systems. That’s the portfolio that convinced Jatt to make the jump to trucking.
“The biggest difference between Waymo Via and Waymo One” — the name of the robotaxi service — “is that Via is a business-to-business product, not a consumer product,” Jatt said.
That means working closely with partners to assess their needs and they really know what they want.
“What’s cool is that there are companies that have driven thousands of trucks for decades,” he said.
An $800-billion business that transports 11 billion tons of freight every year
The scale of the opportunity is vast. In the US alone, trucking is an $800 billion business, moving almost 11 billion tons of stuff around the country every year. Much of that traffic is on highways, avoiding the exceptionally difficult challenge facing autonomous ride-hail services: dense urban environments, which are a business necessity for ride-hailing since that’s where riders are concentrated.
A gaggle of autonomous truck companies has arrived in the past half-decade (and one, Uber-owned Otto, folded in 2018). The biggest names are Daimler, which partnered with Waymo in 2020 to develop a vehicle, Tesla, and Waymo itself. Startups such as Embark and TuSimple remain at the prototyping stage. Tesla’s much-anticipated Semi has brought in preorders from the likes of PepsiCo and Walmart, but the truck, revealed in 2017, hasn’t yet entered production, so Waymo is arguably the front-runner at this stage.
Nicole Gavel, the startup’s head of business development and strategic partnerships, said a Waymo fundamental is to have options when it comes to deploying technology while not introducing so much flexibility that Waymo One and Waymo Via could lose their unified focus.
The Harvard graduate and one-time hopeful for the US Olympic rowing team has an extensive background in retail, including the grocery business, The Gap, Lyft, and the Google mothership working on the shopping applications. She joined Waymo less than a year ago.
“When you have autonomous, you ask, ‘What are some things it can handle in a differentiated way?'” she said. “We can leverage the Waymo driver, and when you think about it from a fleet perspective, it gets really interesting.”
She cited smoothing out the natural peaks and valleys of freight and delivery operations as one example. The potential 24/7 nature of autonomous operations enables her to think about filling those demand gaps, creating savings from efficiency. The pitch has appealed to giants such as UPS and AutoNation, the auto parts chain that’s both a Waymo client and investor. AutoNation joined a venture capitalist, Canadian supplier Magna International, and Alphabet on a $2.25 billion funding round last March.
“I think we’re just getting started, and there’s value in having true strategic partners,” Gavel said. “It’s important to get feedback.”
For Gavel, that means getting relationships to work and work well.
“You want to have that be right,” she said. “It’s sort of like, you don’t want incremental safety, you want full safety.”
It’s surprisingly easy to jackknife a truck
Alphabet is among the world’s premier technology companies, so a significant part of Waymo’s pitch to consumers and businesses is that it’s bringing the full power of Mountain View to bear on a problem that’s been described by some as harder than putting a person on the Moon.
This is where Boris Sofman, head of engineering, comes in. The Carnegie-Mellon doctorate has a background in robotics and artificial intelligence, so he’s literally perfecting the driver that Jatt and Gavel are working to sell to partners. (Sofman also started a company that folded shortly before he joined Waymo in 2019 to work on the R&D project that became Via.)
He stressed that there’s a deceptively large chasm between what looks like a great demo and producing a system like the Waymo driver that can actually move a massive tractor-trailer down a major freeway.
“It’s actually pretty hard to flip or roll a car,” he said. “But it’s surprisingly easy to jackknife a truck.”
He added that a wide range of different thinking is needed when you tackle the autonomous truck problem. For example, it’s one thing to prepare a Chrysler Pacifica minivan to stop itself in urban traffic, but quite another to slow an 18-wheeler on the highway.
“Our mentality has been to leverage everything we can possibly leverage,” he said.
Along the way, some unexpected discoveries have occurred.
“If you want to have a ride-hail service in San Francisco, you have to solve for the entire city,” he said. “In trucking, we have a lot more control over circumstances than we might have on a surface street. On a freeway, you don’t expect to see the chaos.”
That’s led Sofman and his team to consider solving the highway first, using transfer hubs to bring freight into cities on vehicles that don’t have to be autonomous. A useful analogy is to enormous, oceangoing container ships arriving at a port and then being piloted in by a local professional who knows how to navigate the channels and dock the boat without damaging it.
Sofman and his team’s efforts leave Jatt to propose that solution as part of a commercialization strategy — a patient commercialization strategy.
“The product isn’t going to be everything,” he said. “It’s going to be thoughtful. We get people in early and they get excited, but we’re being conscientious about not overpromising. We think it will go gangbusters, and that’s why we’re in it, but it will be slow and gradual.”
As for collaborations — something that Waymonauts are accustomed to — Sofman, Jatt, and Gavel are in the early days of what they all think could be a long engagement. For the moment, the questions they ask each other are simple.
“We’ll eventually have a more systematic process,” he said. “But right now, what I’d ask Boris is, ‘Can you make the truck drive itself?’