America's car industry dismissed the potential of autonomous
driving for years as tech companies plunged ahead. Now Detroit is
racing to catch up. A former auto executive explains the U-turn.
America's car industry dismissed the potential of autonomous
driving for years as tech companies plunged ahead. Now Detroit is
racing to catch up.
By Lawrence D. Burns
This article is being republished as part of our daily
reproduction of WSJ.com articles that also appeared in the U.S.
print edition of The Wall Street Journal (August 18, 2018).
Early in 2011, two top engineers for Google traveled together to
Detroit on what amounted to a diplomatic mission. They had just
spent 18 months on a top-secret project called Chauffeur: the
development of a car that could drive itself over 10 different
100-mile routes on public roads. Now they were looking for a
partner to carry the project forward. "The idea was, if you're
going to make self-driving cars, you have to work with a car
company," recalls Chris Urmson, who made the trip with fellow
engineer Anthony Levandowski. "Maybe they'll sell us cars to build
a fleet. Maybe we're going to be retrofitting our stuff onto their
cars to sell."
But they couldn't find any takers. In meetings with a prime
parts supplier to the car makers and then with the senior
leadership of a major auto company, the pair gave presentations on
their vehicle's capabilities, the number of miles it had driven and
the broad strokes of how their self-driving software saw the road.
The reaction, they say, was utter disinterest -- and dismay that
they were experimenting on public roads rather than on a test
track. "Self-driving technology didn't make sense to them," Mr.
Urmson says. "And it seemed so far out of the playbook that it
wasn't even addressable." As they headed back to the airport, Mr.
Urmson said to his partner, "Well, I guess we're not working with
those guys."
Today, self-driving technology is spurring Detroit's biggest
deals. The SoftBank Vision Fund announced plans this spring to
invest $2.25 billion in GM's self-driving subsidiary, Cruise
Automation -- with GM committing another $1.1 billion of its own
money. In July, Ford announced plans to invest $4 billion in its
own autonomous car startup. This month, Morgan Stanley analyst
Brian Nowak speculated that the company born in late 2016 from
Google's self-driving car team, known as Waymo, could be worth $175
billion -- 40% more than the combined market capitalization of GM,
Ford and Fiat-Chrysler.
Autonomous technology is so hot today that it's easy to forget
Detroit's initial resistance. Why was the center of the American
car industry so dismissive back in 2011, and how did it come around
to the fast-approaching revolution in car travel?
When I joined Chauffeur as a consultant at the beginning of
2011, I became the team's first insider with experience as a
car-company executive. Chauffeur was trying to reinvent the auto
industry, something I'd been trying to do in my own way as the
chief of research and development at General Motors for a decade,
including work on autonomous vehicles. The closest we got during my
tenure at GM was a joint project developing a two-person autonomous
pod with Doug Field, then of Segway (and now making news for his
leap from Tesla to Apple's autonomous car project). Our prototype
made its debut the week after GM CEO Rick Wagoner's 2009
resignation and left little public mark as my employer lurched
toward bankruptcy.
Amid the worst recession that the auto makers had ever faced, I
understood why they would steer clear of ushering in a technology
poised to make them irrelevant. But the deeper reason the auto
companies were late to the revolution is that they mistakenly
believed that their business was manufacturing and selling cars.
They failed to see that their success had always been based on
something more fundamental: helping people to get from one place to
another.
Auto executives initially dismissed self-driving cars in part
because they didn't understand the full potential of digital
technology. But it was also because they were primarily focused on
delivering attractive vehicles to dealer showrooms rather than on
providing compelling transportation experiences to customers.
Detroit was held captive by a century-old business model.
The disconnect on autonomous cars was not only the fault of the
auto industry. Early on, I was struck by the disregard the
Chauffeur team displayed toward Detroit. Chauffeur's engineers
asked me about Detroit's product development cycles. When I told
them that auto companies typically took around three years to
develop a new car, they were astonished. Three years? What on earth
took so long?
The engineers' general attitude was that the auto companies were
lazy, out of touch. They didn't know how to do innovation -- at
least, not the kind that might spur social disruption, the kind in
which Silicon Valley liked to think it specialized. Many on
Chauffeur's team believed that Henry Ford was a remarkable
innovator but that somewhere along the line, that spirit had
withered in Detroit.
The low point in the mutual aversion between Silicon Valley and
Detroit came more than a year after Mr. Urmson and Mr.
Levandowski's failed trip, when Mr. Urmson invited a senior GM
representative to come out to Google headquarters in Mountain View,
Calif. to take a ride in one of Chauffeur's self-driving vehicles.
The executive spent the ride sharing his negative take on the
experience. "I'm sorry," said the guy, according to Mr. Urmson.
"But I just don't get the point. "
Things began to change with the next stage of development, the
combination of driverless tech with another new disruption:
on-demand ridesharing services. Mr. Urmson initially revealed the
idea for a new concept vehicle called Firefly at an all-hands
Chauffeur meeting in December 2012.
Ridesharing was a hot trend in tech investing that year. An
entrepreneur named Sunil Paul had begun arranging rides in San
Francisco via a mobile app called Sidecar, spurring a pair of
mobility entrepreneurs named Logan Green and John Zimmer to roll
out their own anyone-to-anyone ride-sharing app, which they
referred to as Lyft. Uber followed soon after with UberX.
Chauffeur's chief engineer told the team that he wanted to
pursue on-demand mobility as a business model -- essentially, a
driverless version of Uber or Lyft. To do that, Mr. Urmson wanted
Chauffeur to design a vehicle for the express use of ride-sharing
services. He envisioned a world of driverless taxis zipping about
cities, picking up passengers, providing rides, then setting off on
the next call -- what was referred to inside Chauffeur as
"transportation as a service."
I was thrilled with Mr. Urmson's new project. Research that I
had recently led at Columbia University and presented to Chauffeur
concluded that this new model could provide mobility experiences
better than personal car ownership. The point-to-point travel would
be just as fast while allowing riders to forget about
inconveniences such as finding a parking space and refueling.
Our analysis later showed that the new business could do all
that while saving people most of what they paid for trips in
gas-powered, personally owned vehicles, costing them just 20 cents
a mile on average compared with a 65-cent average for drives today.
(Other studies have found similar costs and savings.) This didn't
even count another 85 cents' worth per mile of productive time lost
while driving, which they could use for other things while
traveling as passengers. If just 10% of driving were diverted to
this model, the analysis suggested, it could save on the order of
$150 billion a year in operating costs and about another $250
billion in lost driver productivity.
The new project set Chauffeur on a fascinating design exercise.
What should a driverless vehicle built expressly for the new
mobility services look like? The Firefly designers came up with a
simple, clean and fun aesthetic. Because Mr. Urmson hoped it would
liberate transportation for those who couldn't drive -- the elderly
or disabled, for example -- the vehicle needed to be easy to enter
and exit. To achieve that, the floor would be flat and not too far
from the ground. When the conversation turned to the placement and
feel of traditional car components, the team made a radical
decision: no steering wheel. Why did the Google mobility pod need
one?
In May 2014, Sergey Brin unveiled Firefly publicly. By October,
Uber CEO Travis Kalanick realized how urgent it was for Uber to
develop its own strategy, according to court documents from this
year's trade-secrets lawsuit between Uber and the Google spinoff
Waymo, where I remain a consultant. Mr. Kalanick knew that a
ridesharing business that operated driverlessly could provide its
services for much less than a human-operated rival; the human
driver accounted for a reported 70% to 90% of Uber's cost per mile.
Google had earlier invested $258 million in Uber and placed its
chief legal officer, David Drummond, on Uber's board of directors.
After a board meeting, according to Mr. Kalanick, Mr. Drummond told
him that Google was intending to compete with Uber in the
ride-sharing space, and they agreed that he should recuse himself
from the Uber board.
Soon after, Mr. Kalanick assigned his chief product officer,
Jeff Holden, to develop Uber's self-driving capability -- to
basically create the ride-sharing giant's own version of the
Chauffeur project -- and as quickly as possible.
Mr. Holden identified the world's single greatest concentration
of self-driving brainpower outside of Mountain View: Carnegie
Mellon University's National Robotics Engineering Center in
Pittsburgh, Mr. Urmson's onetime employer. According to The Wall
Street Journal's reporting, Uber offered compensation packages that
included signing bonuses in the hundreds of thousands of dollars
and salaries at least double what the scientists and engineers had
made at NREC. All told, 40 NREC staff would leave. Uber essentially
gutted the place. "I've never seen anything like it," marveled one
Carnegie Mellon observer. "People have been complaining for years
that no one understands how important this technology is. Then Uber
came in and people were like, 'Wow, this thing is real.'"
Uber's mass hiring of NREC's self-driving talent created an
enormous amount of discussion in the auto industry. News stories
about the deal pinged around the email accounts of Detroit
executives. The move amounted to a high-stakes endorsement of
self-driving technology, and it triggered a stampede in the
industry.
Uber would soon become more valuable than General Motors. Google
was one of the world's most valuable companies (there were quarters
when it could have bought GM outright with its cash reserves). The
fact that they were both gunning hard to provide driverless
mobility finally helped to convince the auto industry that it had
to get serious about autonomous cars.
Renault-Nissan CEO Carlos Ghosn announced plans to sell ten new
autonomous vehicles before 2020. Toyota's Akio Toyoda, who had long
opposed the technology, announced a plan to invest a billion
dollars to start a 200-researcher artificial intelligence lab in
Silicon Valley and promised to have vehicles driving themselves on
highways by 2020.
Perhaps the biggest and most startling about-face came from GM.
"I believe we'll see more change in the automotive industry in the
next five to ten years than in the past fifty," said GM CEO Mary
Barra in a December 2015 essay published on LinkedIn. "I have
committed that we will lead the transformation of our industry."
The essay's headline proclaimed 2016 as "the year Detroit takes on
Silicon Valley."
The next month, the company invested $500 million in the
ride-hailing company Lyft. It launched a Zipcar competitor, Maven.
And that March, GM announced its purchase, for $581 million plus
incentives, of a forty-employee Silicon Valley startup called
Cruise Automation, which had begun its self-driving development
more than four years after the founding of Google's self-driving
car project.
As for Waymo, the company retired its groundbreaking Firefly
last summer to focus on what it had set out to explore in the first
place: installing software for autonomous driving in car makers'
mass-produced vehicles. This spring, it signed agreements for such
projects with Fiat Chrysler and Tata Motors' Jaguar brand.
For years, the technologists working on self-driving cars were
frustrated that society at large and the auto industry in
particular were either unwilling or unable to understand what was
possible. Few of us feel that way today. This last decade has been
a learning process for everyone. We've all grown and changed our
minds about how the new technology will develop.
And the learning has happened on all sides. Detroit has been
forced to become more receptive to new ideas. Silicon Valley has
come to recognize that innovation, particularly the world-changing
kind, is a difficult thing to hurry. It requires great leaps of
deduction, constancy of purpose and the discipline to let ideas
mature until they're ready for the public.
One of the most important decisions on Chauffeur's road to
commercialization was who to select to run the company. To fill the
CEO position, Larry Page and Sergey Brin tapped former Hyundai
Motor America president John Krafcik. The hiring of an auto
executive to lead the company that would become Waymo was a
remarkable moment in the history of self-driving cars.
The Chauffeur team had long been untroubled by its lack of auto
industry experience. In many instances, ignorance of the way
Detroit did things was considered an asset. But Mr. Krafcik was an
auto industry guy, and his installment as Waymo's leader was, to my
mind, the first acknowledgment that this quest didn't have to pit
Silicon Valley against Detroit -- that both sides had expertise to
contribute to bringing about the great disruption ahead in
mobility. Mr. Krafcik's hiring was a savvy concession on the part
of Google's founders that maybe, just maybe, the two sides needed
one another.
--This is adapted from Mr. Burns's book "Autonomy: The Quest To
Build The Driverless Car -- And How It Will Reshape Our World,"
co-authored with Christopher Shulgan. It will be published Aug. 28
by Ecco.
(END) Dow Jones Newswires
August 18, 2018 02:47 ET (06:47 GMT)
Copyright (c) 2018 Dow Jones & Company, Inc.
General Motors (NYSE:GM)
Historical Stock Chart
From Apr 2024 to May 2024
General Motors (NYSE:GM)
Historical Stock Chart
From May 2023 to May 2024