Why Waymo Currently Leads the Race for Autonomous Cars
Exploring the company's massive milestone last week, its history, and AV trends more broadly
|John Thomey||Oct 15, 2020||2|
Hi everyone - happy Urban Tech Thursday! The best day of the week for people who love cities and tech. I've been pretty heads down working on some things for Urban Tech that I'm pumped to roll out in the next month or two. It's an exciting time to be thinking about the intersection of cities and the innovation economy.
In today's edition, I'll explore a significant moment in the race for autonomous vehicles that happened last week. Waymo, a subsidiary of Alphabet Inc. (the parent company of Google), announced the opening of its fully driverless service to the public in its Phoenix market.
Quick Waymo background in case you aren't familiar with it. This is how a spokesperson for Waymo described it to me:
"we're developing a fully self-driving level 4 stack, which means the Waymo Driver (our core technology) is responsible for the entire driving task at all times, with no human required in the loop."
Think of Waymo's product as a tech stack, hardware and software, that combines with a car to make it autonomous.
For the rest of the autonomous vehicle (AV) industry, there are companies working on specific components of the tech stack (mapping, sensors, machine learning), pursuing driver assistance (Levels 1-3 on the scale I'll review later) where a human is still required to monitor the system and take over as needed, or limited use cases (closed roadways, fixed routes, low speeds).
Waymo's news is a significant milestone in the arms race for AVs, one of the most aspirational sectors in Silicon Valley.
The size of the AV market is tricky to estimate. The best numbers I found comes from the Automotive & Assembly team at Mckinsey. They estimate by 2040, autonomous vehicles could generate, "a market revenue of $1.1 trillion from mobility services and $0.9 trillion from sales of autonomous vehicles by that year."
Despite the weight of the Waymo news, it felt to me under-discussed in tech and media circles. For reference, arguably the two most important U.S. news outlets, The Wall Street Journal and The New York Times, didn't publish stories on the news. The Washington Post, which has grown its tech team considerably of late, did cover the news.
To be fair, these publications typically cover AVs from a business perspective, and we are still far from understanding the long-term business consequences of the news. Although, I wasn't the only one who felt this moment didn't get the attention it deserved. Platformer founder and longtime tech journalist Casey Newton made a similar comment on Twitter.
It's entirely fair why this news could have gotten lost in the shuffle. We are in the midst of the most critical election of our lifetime and a global pandemic. That said, Urban Tech is the place where we give moments like the Waymo announcement the attention it deserves.
For this piece, I'm going to cover a few things:
Provide some essential context to keep in mind when looking at the AV market
Explore Waymo's fascinating history within Google
Explain what Waymo announced last week, why it's important, and some reasons it might not have gotten the attention it deserved
A couple of small asks before diving in:
Please continue to share Urban Tech with folks you think would be interested! We are continuing to grow rapidly, and word of mouth is still the best method for our growth at this stage. We all have a vested interest in the success of our cities.
We launched the Urban Tech Instagram page last month. Go follow us!
Okay, let's dive into some helpful context for analyzing the AV market and new developments.
Rules of the Road
A few issues I see with conversations about AVs and industry developments:
The oversimplification of how hard a problem building a driverless car is
The oversimplification of the levels of AVs
To the first point, people have dreamed of cars without drivers since the early 20th century. We are just now at a point where we have the technology to develop the solution. We needed significant leaps in several fields, including computing, physics, manufacturing, mathematics, and many other disciplines. Having a driverless car will result from thousands, if not millions of people, advancing their fields. It can be easy to miss how big an endeavor it is to create a fully driverless car.
My second pet peeve for AV discussions is treating all autonomy levels as equivalent and not distinguishing between the levels.
By all accounts, the AV market right now is probably between stages 3-4. If Waymo One delivers on its promises, this would be the best example of stage 4 autonomy (and remember it's being commercialized right now in a big metro). Stage 4 is also how Waymo describes its current product.
Tesla's Autopilot features are generally considered to be level 2 or level 3 because you are required to keep your hands on the wheel at all times. Elon Musk, a man who knows how to get people hyped, claimed in June Tesla is on the verge of level 5 autonomy.
Musk also had comments on the Waymo news, and the team at Waymo cleverly shot back.
The former communications consultant in me thinks this is a pretty remarkable comeback for an AV brand against a competitor like Musk, who dominates the automotive conversation.
It’s always important to keep top of mind how many companies are working on AVs. This chart from 2016 gives you somewhat of an idea of the wide variety of companies in the space.
There are a ton of companies working on AV technology. Most you’ve never heard of because they work on super-specific hardware or software that’s part of the AV stack. A lot of them have millions in VC backing too.
A final point for this intro: The major players in AV tech have a pretty incestuous relationship with each other by virtue of investments, partnerships, and M&A.
The below diagram from CB Insights gives you an idea of how complicated the ecosystem is. Note: Amazon is in the AV market now too after acquiring ZOOX over the Summer.
(I’d recommend bookmarking this diagram for quick reference to understand the complicated autonomous driving partnership ecosystem.)
Okay, time to dive into Waymo.
For this section, I leaned heavily on the incredible reporting out there about Waymo, Uber, and AVs in general. Specifically, Mike Isaac's book on Uber, Super Pumped, is an excellent resource for understanding transportation tech in Silicon Valley. Coverage from Aarian Marshall and Alex Davies at WIRED, and Daisuke Wakabayashi at the New York Times was also super helpful for this section and understanding Waymo’s history.
Waymo started like several bold ideas in tech over the last few decades have: as an internal project within Google. More specifically, the project began in Google's moonshot factory, X, in 2009.
As I mentioned in my Waze piece (Google's crowdsourced navigation platform), the company, and its founders Larry Page and Sergey Brin, have ambitions in tech that impacts cities — particularly transportation tech. Sergey Brin has a particularly strong interest. You might have heard of his flying car venture Kitty Hawk. Larry Page also is passionate about urban tech, having led Alphabet during the formation of Sidewalk Labs.
Two engineers, Sebastian Thun and Anthony Levandowski (a lot more on him in a bit!), are credited with launching the self-driving car program within X. Thun is currently the CEO of Kitty Hawk, the flying car venture. Both the engineers and many of the original project's team members participated in early self-driving car competitions and challenges like the DARPA Grand Challenge, an autonomous vehicle competition aimed at advancing the technology for military applications. Levandowski even sold an AV startup he built in his spare time to Google before working on AV projects within X.
The project didn't start with executing autonomous vehicle plans — in the early days they were focused on mapping and other key technologies that AVs require to operate. By all accounts, the team quickly pushed Google leadership to begin the crucial step for AV development, open-road testing.
Open-road testing is the only way to train your software to drive on its own eventually.
This quote from Super Pumped encapsulates the engineer's technocratic vision for creating AV technology:
“For years, Levandowski believed that humans moved around the world in a way that made no sense. Tens of thousands of people were killed in automobile accidents annually. Traffic in major urban areas, especially the San Francisco Bay Area, was abysmal. People clogged the streets with inefficiently operated cars. One person to every vehicle on the road was inefficient and wasteful. A fleet of self-driving cars, used only when necessary, would be far cleaner and more cost-effective.”
I also really like this explanation because these principles are the same reasons driving Tesla, Cruise, Uber, and hundreds of other companies to spend on developing the technology.
It took Google sometime before it was comfortable with the idea of open-road testing. The company was running a lot of road tests undetected for almost 2 years before being detected. In a wild story recounted in Super Pumped and other Waymo coverage, Levandowski became so frustrated with the pace of testing he commissioned a lobbyist to sway AV testing regulations in Nevada, without Google's knowledge.
Tensions continued to grow between Levandowski and leadership. Levandowski felt handcuffed, thinking he should be running the project entirely and grew frustrated with the development pace.
As the project grew closer to a viable commercial product, Google brought in former Honda Motor America CEO to lead the project.
In early 2016, Levandowski eventually left Google to pursue other ventures (hold on to this thought).
2016 proved to be a pivotal year for Google solidifying itself in the race for AVs. In late 2016, Google spun the self-driving project into a real company — Waymo. The name was picked because the company brought "a new way forward in mobility".
The formation of Waymo as a company was Google saying its research project was ready for commercialization. Before being spun out, the project was protected from any financial scrutiny or pressures to make money. Simultaneously, some of the other major tech players of the last decade were also developing AV technology: Tesla, Uber, Apple, Lyft, — not to mention traditional automotive players like BMW and Mercedes.
The big Uber-Waymo Scandal
After Levandowski left, he founded a new AV company, Otto, which was quickly acquired by Uber for $600 million.
Google was pretty skeptical about this situation and eventually found Levandowski stole crucial intellectual property:
Levandowski downloaded more than 14,000 confidential files related to the self-driving program from Google’s servers directly to his personal laptop. Among the files were designs for Waymo’s proprietary lidar circuit boards, one of the crucial components necessary for most self-driving cars to function. After downloading the files, he copied and transferred all of the 9.7 gigabytes of Waymo data over to a personal, external hard drive. When he finished, Levandowski installed a new operating system, erasing the contents of his work laptop hard drive. “After Levandowski wiped this laptop,” Waymo’s attorneys would later say, “he only used it for a few minutes, and then inexplicably never used it again.
My guess: the resulting legal scandal that erupted between Waymo and Uber is likely the first time you heard about Waymo. Google was even hesitant to take legal action against Uber and Levandowski.
Investigators may not have pieced all of it together, were it not for a mistake from one of Waymo’s own lidar component suppliers. In February 2017, months after the Otto acquisition, the manufacturer accidentally included a Waymo employee on an email, an email which happened to include a schematic of a component from Uber’s most recent lidar design. The Waymo engineer noticed something peculiar; Uber’s lidar component looked like a carbon copy of Waymo’s hardware.
Oops. All bets were off after that, and Google put its weight behind a multi-year legal battle that it won. For his part, Levandowski pleaded guilty in the criminal case and was just in August sentenced to 18 months in prison and had to declare bankruptcy.
Uber for its part paid a costly price. Uber settled in 2018 and agreed to pay Waymo $245 million worth of its own shares. This is a massive price, but Uber’s new CEO Dara Khosrowshahi needed to move the company past the scandal.
Interestingly, this means Uber is now incredibly invested in the success of Waymo’s AV tech. It’s highly likely that if Uber deploys AVs in its fleet, it will be using Waymo technology because of all this. Uber is still working on autonomous within its walls, but smart folks like Ben Thompson believe Waymo could be squeezing away Uber’s dreams of developing its own tech.
this should be the final (final!) nail in the coffin of Uber’s ongoing self-driving boondoggle, which has been a massive waste of time, money, and goodwill. I have been predicting all along that Google was never going to build a competitive ride-sharing service, but would rather end up licensing Waymo tech, which would benefit Uber most of all. That seems even more likely to happen now that Waymo has an actual profit motive.
For the ridesharing model to reach its promise, at some point Uber or another player, needs to figure out how to deploy AVs to remove the biggest operating cost for ridesharing, the driver.
Even during the Levandowski scandal, under the leadership of John Krafcik, Waymo continued its momentum while also reaching key goals that continue to solidify its business model:
The company reached a partnership deal with Chrysler and is its “exclusive partner for future Level 4 autonomous vehicles”.
Waymo in 2018 launched Waymo One testing in the Phoenix metro area. Waymo also has conducted testing in places like Atlanta, Austin, Detroit, and Kirkland, WA.
It launched a trucking solution called Waymo Via that uses its proprietary tech stack.
It also raised external funding twice in 2020.
The funding timing makes a lot of sense with the COVID timeline. The entire transportation sector has faced significant hurdles to say the least. AV developers have definitely not been insulated. Waymo’s most recent valuation was $30 billion when it raised money. In August of 2018, Morgan Stanley estimated Waymo’s valuation at $175 billion. The entire market has redefined expectations, but the $30 billion valuation is pretty remarkable for a company just beginning to commercialize its product.
The business strategy for Waymo is to build out its Driver (its proprietary tech), and then deploy this tech across multiple use cases and revenue models, including ride-hailing like the Waymo One service, trucking and local delivery like the Waymo Via business unit, and eventually down the line maybe even offer personal AVs to consumers.
Explaining the moment’s significance
I hope by now it’s pretty clear why the news last week from Waymo matters. And I hope by the end of this piece you’ll agree with me it deserved way more attention.
Here’s the main piece of news from Waymo CEO John Krafcik’s blog post:
Beginning today, October 8, we’re excited to open up our fully driverless offering to Waymo One riders. Members of the public service can now take friends and family along on their rides and share their experience with the world. We’ll start with those who are already a part of Waymo One and, over the next several weeks, welcome more people directly into the service through our app
Check out this video Waymo posted for what Waymo One looks like in the wild. It feels like the stuff of science fiction, but it’s happening right now in Arizona:
So, why could the news have been lost a bit?
We are in the middle of the highest-velocity news cycle ever between the election, the pandemic, and any of the other crazy things happening. Once 2020 is a few decades behind us, I believe we will realize that we missed recognizing some historical events in the moment because of everything else's noise.
Another big reason I think the news didn't splash as big is progress for AV development has been so incremental. Rather than one major scientific advancement, lots of little moments have added up to the Waymo One news. This theme might have made the moment last week feel less impressive. People are also a little burnt out on hearing about AVs.
Time will help us understand the overall business implications of the news for Waymo. This piece from a Financial Times story on Waymo’s latest valuation was very interesting to me in the context of Waymo’s long term play:
Narrowing its focus on the core technology could turn Waymo into a pure tech company offering a range of services to carmakers, said Mike Ramsey, an analyst at Gartner. That could make its business more akin to the massive cloud computing platforms that Google and a handful of others have built, he added, working in the background to supply much of the computing power and AI on which other businesses depend.
This strategy would mean slower revenue growth at higher margins, rather than fast revenues at low margins, like running a consumer ridesharing service. More and more, this looks like the approach Waymo will take long-term as a business, but the company is applying its tech to a variety of use cases and business models — which is another reason Waymo is so interesting to me.
There is also a ton of unpredictable value, through more data and better insights, that could be derived from network effects as Waymo tech scales. A licensing model would allow for that to happen even faster.
The Waymo One announcement is the kind of moment we eventually see movies made. Krafcik in his blog post talks about one of Waymo's first major public milestones from 2015. In Austin, a blind man took the world's first ride in a self-driving car on public roads. Some commentators compared the event to Kitty Hawk.
To me, the news last week feels at least 10x bigger and is the kind of moment that propels the industry forward. While the long-term implications remain to be seen, at the very least, last week's news deserves to be a chapter in the inevitable history book of AV technology.
What I’m Reading This Week:
The Wall Street Journal: How Airbnb Pulled Back From the Brink
Catch the huge Airbnb profile in the Journal? This is what I said last week about Airbnb’s press strategy going into the IPO:
“At points like in 2018, Airbnb has been profitable. Leading up to the IPO, a story about how it can get there again relatively quickly is what the company will be looking to land.”
Airbnb landed that story with this Wall Street Journal piece. Some bits from it:
“In August, more than half of bookings made were for stays within 300 miles of the guest’s location, according to the company.”
“The upswing has put the home-sharing giant on a path to go public and report a third-quarter profit this year, according to investors, something that seemed all but impossible months ago.”
“With business recovering, Mr. Chesky wanted to send a strong signal to investors and boost morale among employees, some of whom were anxious about their stock options expiring at the end of the year, according to the person Mr. Chesky has frequently consulted.”
Now, Airbnb still needs to continue to sell its business model to investors leading up to the IPO, but a story like this helps bolster the argument.
Sidewalk Labs: Announcing Delve: Discovering radically better urban designs
Google urban innovation company Sidewalk Labs announced a new tool for design called Delve.
“Delve uses machine learning to help development teams discover the best neighborhood design for their project, based on the priority outcomes that matter most. By revealing the optimal design option, Delve helps development teams exceed their project economic goals while improving quality-of-life outcomes for residents and businesses.”
“This discovery process begins with a development team using Delve to identify a project’s planning inputs (such as gross floor area for commercial or residential use), site constraints (such as height limits), and priority outcomes (such as cost or daylight access). Delve’s intuitive interface makes this initial step easy, with support available from a computational designer who is a trained architect.”
“From there, Delve empowers a development team to get from endless possibilities to one optimal design. Teams can update their priorities and planning inputs as stakeholder feedback changes, exploring new possibilities in minutes instead of months. Delve then automatically finds the highest-performing options based on a project’s priority outcomes — leading to approvable and financially feasible neighborhood designs that exceed stakeholder expectations.”
I’m super excited to see how projects and developers leverage Delve. Check out the full details for the tool in the blog post from Sidewalk Labs.
[Disclosure: at my first job out of college, working as a consultant for BerlinRosen, Sidewalk Labs was one of my first clients.]
A company helping digitize the mortgage process raised a new funding round.
“Snapdocs, which is used by some 130,000 real estate professionals to digitally manage the mortgage process and other paperwork and stages related to buying a home, has raised $60 million in new equity funding on the heels of a few bullish months of business.”
“In August 2020 — a peak in home sales in the U.S., reaching their highest level in 14 years — the startup saw 170,000 home sales, totaling some $50 million in transactions, closed on its platform. This accounted for almost 15% of all deals done that month in the U.S. Snapdocs is now on track to close 1.5 million deals this year, double its 2019 volume.”
“The Series C is being led by YC Continuity (Snapdocs was part of Y Combinator’s Winter 2014 cohort), with existing investors Sequoia Capital, F-Prime Capital and Founders Fund, and new backers Lachy Groom (formerly of Stripe and now a prolific investor), Maverick Ventures, and DocuSign, a strategic backer, also participating.”
Thanks, everyone for reading this week’s edition! Remember to give this post a share. Talk to you next week.