Zoox CEO Aicha Evans keeps the self-driving robotaxi faith
In January 2019, Aicha Evans sat on a Jamaican beach staring at a career crossroads: Should she remain Intel’s chief strategist or become chief executive of Zoox, a self-driving-car startup?
Born in Senegal, raised in Paris, and educated at George Washington University, Evans, then 49, weighed her options. She’d spent two decades rising through the ranks of semiconductor companies—Skyworks, Rockwell Semiconductor, Conexant, and then Intel. But after a dozen years at America’s most iconic chipmaker, “I had decided that I was not going to leave to go to another big company,” Evans recalls when we chat in November, “because at the end of the day, it was sort of going to be the same everywhere. I already knew the movie, the plot, the actors, the drama.”
Zoox, by contrast, would take Evans into uncharted territory. Not only would she now be joining the startup’s quest to build an autonomous, battery-electric robotaxi and ride-hailing business, but she’d also become the first African-American female CEO of an autonomous vehicle (AV) company. There was a time, early in Evans’s career in tech, when corporate veterans like her were routinely tapped to run startups, but those days have largely passed.
“I am not a Silicon Valley-like founder,” she says. “I take a lot of risk, and I like disruption, but I grew up through the corporate engine.”
Evans decided that if she were going to leave Intel, the position would have to satisfy an improbable range of criteria: a small, private, Silicon Valley-based company, working with impactful technology and helmed by founders “who wanted or needed to partner with somebody like me,” Evans says, “because they truly felt that that’s what they needed and not because someone told them, ‘Okay, you need some adult supervision.’”
So she visited Zoox’s headquarters in Foster City, California, where the startup was building an early prototype for the robotaxi it aims to commercialize. “Instead of just sitting in a conference room, we walked around, so I got to feel the energy. I got to see the vehicle being worked on. Then I went on vacation for two weeks in Jamaica and made the final decision.”
Evans joined Zoox in February 2019, responsible for mobilizing teams across the company’s robotics, vehicle engineering, software, artificial intelligence, and quality assurance departments to create a driverless vehicle from the ground up. She was competing with a bevy of rivals also racing to shape the autonomous future, backed by such behemoths as Google, General Motors, and Ford. Zoox’s ambitious two-part goal: to create a smartphone-based robotaxi service to compete with Uber and Lyft as well as the purpose-built vehicles to supply it.
Autonomous driving represents a colossal challenge poised to reinvent society—eventually. Over the past decade, many of tech’s brightest minds have spent billions of dollars trying to make self-driving cars a reality and will undoubtedly spend billions more on the pursuit. Meanwhile, the industry’s hype cycle has given way to a trough of disillusionment, with several leading evangelists even becoming naysayers. “You’d be hard-pressed to find another industry that’s invested so many dollars in R&D and that has delivered so little,” Anthony Levandowski told Bloomberg in October. Levandowski, cofounder of Google’s self-driving-car program and former poster boy for autonomous technology, reversed his position following his criminal conviction for taking Google’s trade secrets to his next job at Uber. Now, he claims, the tech is only suited to controlled environments, such as the mines that his new autonomous trucking startup, Pronto, serves.
The industry can veer from hope to despair, sometimes in the same day. In October, Israel-based Mobileye completed an IPO—its second in eight years, after being owned by Intel in between—whereas the once-promising AV startup Argo AI abruptly disbanded. Volkswagen and Ford had pledged $3.6 billion in investment but pulled their support as the project lagged. Ford originally expected to roll out autonomous technology across its passenger cars in 2021 but said it’s no longer considered a priority. Amazon also reportedly planned to use Argo’s technology in its fleet of battery-electric Rivian delivery vans.
“Forget about profits,” Levandowski said. “What’s the combined revenue of all the robotaxi, robotruck, robo-whatever companies? Is it a million dollars? Maybe. I think it’s more like zero.”
Ask Evans about these kind of comments, and she says that’s just noise to her. “It’s very important to have irrational belief,” she tells me, “because that’s what gives you the energy to keep pushing forward.” While it is true that Zoox, founded in 2014 at Stanford University, has yet to generate revenue, she’s focused on the long game. She has faced daunting odds before, working on wireless telecommunications projects during her career. “We had this dream 20 years ago of what is now possible today via data and internet in your pocket,” she says. “But let me tell you that when we were working just to have 3G roaming work, putting the internet in your pocket seemed pretty unattainable.”
Zoox is currently testing its autonomous stack—the bundle of lidar, radar, and cameras that help its vehicle perceive and adjust to its surroundings. It’s doing so in Toyota Highlander SUVs across San Francisco, Las Vegas, and Seattle. These trial runs are in service of Zoox’s goal to create its own driverless vehicle from scratch, an endeavor that even Apple has reportedly struggled with, due to its complexity.
Google’s Waymo and General Motors’ Cruise have already launched commercial driverless robotaxi services in San Francisco. But unlike competitors, or “fellow travelers,” as Evans calls them, Zoox isn’t merely retrofitting regular cars with an autonomous stack. Instead, Zoox’s purpose-built robotaxi intends to forego pedals, a steering wheel, a driver’s seat, and a discernible front or rear end.
The four-passenger cube’s symmetry is suited to bidirectional driving, allowing it to navigate narrow streets without needing to execute any treacherous three-point turns. The interior features plenty of space for passengers to work, sleep, or socialize while traveling. “Our approach is different because we are not trying to make the cars of today self-driving,” Evans says.
Zoox’s design features a sensor pod atop each corner of the robotaxi’s flat roof, which allows the vehicle to see more than 300 feet in any direction. “The shape of the vehicle is perfect for autonomous driving because each of those sensor pods has a 270-degree field of view,” Zoox cofounder and CTO Jesse Levinson said during the vehicle’s public debut on a TechCrunch conference stage in May. “Because there’s one on each corner, not only can we see everything 360 degrees but we have an overlapping, redundant field of view that helps us see around things.”
For the most part, the technology needed for a car to get around town on its own already exists. “There’s nothing from a technology standpoint that is missing for us to solve this problem,” Evans says. “But deploying it in a way that’s quantifiable, repeatable, and scalable is going a little slower than I expected.”
One unresolved challenge will be Zoox’s business model. “We don’t sell you a vehicle,” Evans notes. “We sell you a ride, and we amortize it over many riders all day long.”
That said, “if you are using expensive vehicles for many hours a day and have a high operating and maintenance cost on the vehicles, how many people or things do you need to ‘deliver’ to be profitable?” asks Ram Vasudevan, associate professor of mechanical engineering at the University of Michigan. “It’s a simple question to ask, but a hard one to answer.”
The onset of the pandemic in the spring of 2020 obviously did not help, slowing the momentum that autonomous driving efforts had appeared to be gaining at the time. Zoox’s vision of safer, greener, and more enjoyable urban transportation at scale would be relegated to an even more distant future. Evans says that Zoox saw the need for corporate support in order to roll out its technology in cities at scale. “That’s when we started thinking in certain situations an acquisition would not be a bad thing,” she says. “Obviously you make a list of people who can afford it, but it wasn’t like we entered 2020 thinking, ‘Okay, let’s go get bought by Amazon.’ But we started thinking about what do we want to do? Is it realistic in this industry?”
Waymo has Alphabet/Google and Cruise has GM, so now Zoox, too, would have its own goliath: Amazon acquired the company in a $1.3 billion deal in June 2020.
Amazon has been a leader in buying, deploying, and advancing warehouse automation technology, as well as artificial intelligence through its Alexa voice-enabled virtual assistant. Evans says that Amazon has been patient as Zoox navigates a burgeoning industry with no clear horizon, but one must note that the retail and cloud-computing giant has been cutting programs and workers from its budget of late. That includes scaling back testing for its Scout robot, a last-mile delivery service that had been attacking a simpler challenge than scaling autonomous driving.
Evans, though, is undaunted. “I feel zero pressure to make time lines,” she says. “I mean, I feel my own pressure to get to market and scale from there, and that’s what [Amazon would] like, too, but that’s just not the type of relationship we have. They understand AI from other industries, and they are very, very realistic and very, very clear about how long this is going to take and what it’s going to take.”
Machine learning can program a car to handle the majority of driving conditions it’s likely to face, but it cannot account for the infinite permutations of potential traffic and weather patterns. Evans often speaks about “edge cases” that are near impossible to program. Human drivers possess an ability to read nuance that driverless cars can’t. Since it’s impossible to qualify all of the potential situations on the road, algorithms can only generalize how the car should react when faced, for example, with a bicyclist on its right, roadwork on its left, and unexpected debris from a tractor trailer ahead of it.
It can get confusing for a car. General Motors’ Cruise self-driving subsidiary recalled 80 cars this summer to repair a software glitch following a crash in San Francisco that injured two people. An investigation revealed that Cruise’s Chevrolet Bolt AV failed to predict the path of an oncoming vehicle when making a lefthand turn at an intersection.
Google’s Waymo tops the list for autonomous-vehicle crashes, according to a report from the National Highway Traffic Safety Administration, with 62 incidents between July 2021 and May 2022. The data show 23 crashes for Cruise and 12 for Zoox during the same period.
Earlier this month, one of Zoox’s Highlanders rear ended a tractor trailer stopped at a traffic signal in Las Vegas. The company said its SUV was driven by a human operator, who, according to the police report, said she expected the vehicle to stop on its own, even though it was not in autonomous mode when the crash occurred. The incident underscores the potential challenges, including education and driver awareness, in adapting to this new frontier in transportation.
It’s unclear where fully driverless vehicles are headed or when they will arrive, but “there is no doubt in all of our minds that this is how people are going to move around cities in the future,” Evans says.
When asked when the driverless future will definitively arrive, Evans likes to suggest that autonomous driving is one generation away from achieving critical mass. Evans, who has a 17-year-old daughter and 14-year-old son, expects the technology to be widespread enough that her “kids’ kids” won’t need driver’s licenses. Until then, she predicts a slow and steady climb for the industry, with plenty of consolidation and reshuffling.
“This is not a get-rich-quick scheme. We weren’t looking for the biggest valuation. This was not about what’s the best exit you can get and then moving on to the next thing.”
“This is not a job,” Evans adds. “This is a mission.”
(100)
