It is a breezy spring day in Willows, California, and a motley collection of cars is preparing to take on the winding course at the Thunderhill Raceway. But unlike most auto races, this isn’t a test of the skill of the person sitting behind the wheel. These cars are driving themselves.
The entrants in the Self Racing Cars challenge range from navigation technology start-ups and component suppliers to budding software companies and students. In a narrow sense, the race is a failure: after two days of practice, most teams never manage to make it around the course fully autonomously.
Still, there is electricity in the air. Programmers buzzing from energy drinks make tweaks to their codes while investors stroll in the parking lot to check on their companies. Self-driving cars are the hottest thing in Silicon Valley, and this race is a way for the smallest, boldest start-ups to show their stuff.
“You are seeing a Cambrian explosion of different possibilities, as each different start-up explores a slightly different space or path through the problem,” says organiser and investor Joshua Schachter, boldly comparing the proliferation of driverless car start-ups with the appearance of complex animals on earth.
Last year there were just three entrants in the race. This year, there are 10 — just one indicator of the youth and the rapid growth of the driverless sector. Entrepreneurs and investors are rushing to cash in on a trend that has already made several fortunes, and autonomous vehicle start-ups seem to pop up almost every day. Investment in the sector reached an all-time high of $750m in the first quarter of this year, according to CB Insights.
$1bn GM’s investment in self-driving car start-up last year
$680m Uber deal last year to buy Otto, an autonomous trucking company that was less than a year old
$15bn Intel deal to to buy Israel’s Mobileye, which makes self-driving sensors and software
But even enthusiasts are beginning to worry that the sector might be overhyped. Carl Bass, one of the competitors in last month’s race and a Silicon Valley veteran, is among them.
“There is such a crazy thing going on in the market right now around autonomous vehicles,” he says as he hops into his self-driving go-kart. “It is kind of like if you can spell ‘self-driving’ you can sell it for a billion dollars.”
Yet it is not just Silicon Valley money pouring in. The world’s top automakers such as Ford and General Motors have joined Google’s parent Alphabet, Uber and other tech companies in funding research for self-driving technology.
For the automakers, autonomous vehicles pose an existential threat. Instead of owning cars, consumers in the driverless age will simply summon a robotic transportation service to their door. One venture capitalist says auto executives have come to him saying they know they are “screwed”, but just want to know when it will happen.
This desperation has prompted a string of big acquisitions, which in turn has fuelled the hopes of the fortune-seekers in Silicon Valley. Last year GM paid $1bn for Cruise, a self-driving car start-up, while Uber paid $680m for Otto, an autonomous trucking company that was less than a year old. In March Intel spent $15bn to buy Israel’s Mobileye, which makes self-driving sensors and software.
As in past tech hype cycles, the business model for driverless cars is not clear, nor is the timeline for how long it could take the market to develop. At the moment, driverless cars have nearly mastered highway driving but still struggle in complex urban environments, and there are huge legal and regulatory questions to be worked out.
But such details do little to diminish the promise of the technology, say entrepreneurs and investors in the sector. “This is going to be earth-shattering for the [transportation] industry,” says Sebastian Thrun, one of the pioneers of self-driving cars at Google.
“Transportation is a multitrillion dollar industry. I would argue that, given the potential of this technology, we are under-hyping it.”
Other driverless tech evangelists echo this view. “This is almost like something that should be a mission of the human species, instead of a company,” says James Wu, chief executive of mapping start-up DeepMap. “It can benefit everybody and save lives.”
In the near term, one of the biggest challenges for the sector is a severe talent shortage. People with expertise are in high demand, giving them extraordinary leverage. Instead of taking a job with a salary, they launch start-ups, then sell out to a company that wants to hire them. This lucrative route has come to be known as the “acqui-hire”.
The purchase prices of recent acquisitions work out to “roughly a $10m price tag per person,” says Mr Thrun, who is often referred to as the godfather of self-driving cars. “It is a lot of money.”
He hopes the price tag will fall as more engineers gain the necessary skills, noting that the self-driving car seminar he co-teaches at Udacity has had more than 25,000 applicants.
The most controversial acqui-hire was when Uber snatched up a small trucking start-up, Otto, founded by Anthony Levandowski, a former Google engineer. Mr Levandowski was an early member of Google’s self-driving team, now known as Waymo, earning more than $120m in bonuses for his work.
But outside Waymo he was worth even more. Mr Levandowski started discussions with Uber before leaving Waymo; when he founded Otto, it was purchased by Uber for $680m in equity in just six months. (The events that surround Mr Levandowski’s departure are the subject of a lawsuit, which alleges that Uber infringed on Waymo patents and stole trade secrets. Uber has denied wrongdoing).
With headline deals like these, investors such as Amy Gu, a partner at venture capital firm Hemi Ventures, fear that the sector may be attracting the wrong type of entrepreneur. “I think a lot of people are attracted to the industry purely by the capital that is flowing in, instead of by their desire to figure out this problem,” she says.
While she is still investing in autonomous start-ups, she says she seeks companies that have a revenue model rather than just an exit strategy.
Self-driving engineers say the frenzy has complicated life for them too — and poses risks in terms of safety and reputation. US regulators have so far been fairly permissive about testing autonomous vehicles, but many worry that one terrible accident by an overambitious start-up would quickly change the environment. Testing of autonomous vehicles is already legal in more than a dozen states, and federal guidelines were issued last year.
Some of these start-ups operate in a kind of paranoid secrecy, so that their competitors do not know what they are doing, or even who works for them. Zoox, based in Menlo Park in the San Francisco Bay area, has raised hundreds of millions of dollars in venture funding without ever showing its technology or its “robo taxi” design in public.
David Liu, founder of self-driving start-up PlusAI, says he has seen rivals try to draw new employees with the promise of a quick acquisition.
“You have a couple of guys who worked in Tesla or Apple or Google before, they start a company and expect to be sold in six months,” he explains. This mindset is often accompanied by unfounded marketing claims that, he worries, could damage the credibility of the entire industry.
Automobiles are a heavily regulated industry, and carmakers in Detroit often speak a different language to entrepreneurs in Silicon Valley. Established carmakers are terrified of missing out, but also afraid of damaging their brands by moving too quickly. Start-ups, in turn, can be overly dismissive of the carmakers’ expertise.
A widespread assumption is that as autonomous vehicles become accepted, people will stop buying cars altogether and instead use autonomous transportation fleets that they can summon by smartphone. In the tech world, there are three main contenders working on services like these: Waymo, Uber and Tesla. Tesla has already been pushing the boundaries with intelligent driving assistance in its cars. Waymo and Uber are testing robo-taxis, albeit with a human still sitting in the driver’s seat.
It is unclear where the clutch of new autonomous vehicle start-ups will fit in. With no path to the consumer, most are not able to generate revenue, and some are struggling. Several engineers specialising in artificial intelligence, an area that is core to autonomous driving, told the Financial Times they had recently left the sector because of doubts about its viability.
“In the actual gold rush, you knew there was gold out there somewhere, and people were able to mine it,” says Josh Hartung, chief executive of Poly-Sync, which makes software for autonomous vehicles.
In the autonomous gold rush, it’s less obvious whether there is any gold there, he says. “There is effectively zero revenue that is being produced by this industry,” he points out. “You’ve got this massive, multi-billion-dollar science project, that’s basically on VC life support, until such a time as somebody ships and makes money.”
Even when self-driving technology becomes ready, regulation, public acceptance and developing a viable business model will form barriers to widespread implementation. “I think people have underestimated some of the forces that will resist driverless vehicles,” says Mr Bass.
At a time when many politicians are obsessed with saving jobs from robots, a technology that is poised to put millions of drivers out of work has plenty of natural enemies. “Silicon Valley may have underestimated the Teamsters,” he says, referring to the powerful labour union.
Such worries seem almost beside the point at the Self Racing Cars competition. Several attendees compare the race with the Homebrew Computer Club, the hobbyist group where Steve Jobs and Steve Wozniak first came up with their version of the personal computer. Like the advent of the PC, the prize here is potentially vast, promising to revolutionise the way people and goods are transported.
“Everybody’s big dream is, I want to be the next million-dollar or billion-dollar, start-up,” says Anthony Navarro, head of the student self-driving car team from Udacity, which had just six weeks to develop the software for its autonomous car. “There’s opportunities to be your own CEO.”
As the sun starts to hang low in the sky at the racetrack, Mr Navarro and other students gather for a pep talk from George Hotz, 27, the founder ofself-driving software start-up Comma.ai. Although they are bleary-eyed from all-night coding sessions, they listen intently.
As he addresses the group, it is the words on Mr Hotz’s T-shirt that seem to say it all: “We are gonna be so rich,” reads the slogan on the back, alongside his company logo.
1 The scanner on the car emits a pulse of laser light, similar to the way bats emit sound waves for echolocation
2 An object reflects the pulse back to the scanner’s receiver
3 Measuring the time taken for the pulse to return enables distance to be calculated accurately
4 By emitting pulses at a very high rate, thousands of times per second, a very accurate picture of the surrounding environment can be modelled
5 Rotating scanners can yield a 360-degree model, while “solid state” lidars that do not rotate can scan up to 120 degrees
* Light detection and ranging
Get alerts on Driverless vehicles when a new story is published