Robotaxis haven't taken over the streets, but that doesn't mean self-driving cars are all hype

Published Sat, Apr 30 20229:19 AM EDT

It wasn't supposed to be this hard. Morgan Stanley, in a lengthy 2013 report, told investors that self-driving taxis are "the new auto industry paradigm" and assured readers that "completely autonomous cars are set to available before the end of the decade." But the decade has ended and robotaxis aren't a common sight on American streets. Amid the hype, it's easy to miss that the billions that poured into the sector haven't been wasted. Investors are already reaping the rewards of these efforts. The team of Morgan Stanley analysts behind the pivotal report, which included the now widely followed Adam Jonas, argued that most of the technology required to make fully self-driving vehicles was already available, with "only incremental R & D required" to bring it to market. In fact, as they saw it at the time, the main obstacles to widespread deployment of self-driving vehicles were likely to be regulatory. "We believe autonomous vehicle technology is a smaller leap than full electric vehicles," the team wrote, arguing that then-new innovations from smartphone development (like miniaturized cameras) would help speed development of driverless vehicles. EVs, on the other hand, would require "unknown battery breakthroughs in a lab or significant macro disruption" to make them viable as more than niche products. There have been ten years of immense investment and hype cycle around robotaxis. That investment has educated a generation of engineers that have now gone and found the less hyped, but easier to implement, applications across all kinds of automation. CEO, Ouster Angus Pacala And, of course, they argued that the advent of self-driving cars would present "an existential threat" to global automakers as the real value would come from (and be realized by) software providers — just as with smartphones — leaving automakers to become "low-cost assemblers" of connected, autonomous vehicles. The point here isn't to make fun of Morgan Stanley. In retrospect, that report was a well-thought-out attempt to guide investors through the hype around self-driving technology that was so prevalent at the time. But as we all know, things haven't unfolded quite the way the Morgan Stanley team (and to be fair, many others) expected. Here's a look at where things stand in the robotaxi race now, and at how investors might best think about opportunities and challenges in this still-developing space. It wasn't just hype There's a case to be made that all the hype may have been justified because the business opportunity for self-driving technology could be huge. Estimates of the size of the opportunity have varied over the years, but nearly all have had big, big numbers. For starters, that Morgan Stanley report estimated that autonomous vehicles "can contribute $1.3 trillion in annual savings to the U.S. economy alone, with global savings estimated at over $5.6 trillion." Those were huge numbers, but at least some analysts thought they might be too conservative. Strategy Analytics, in a report commissioned by chip giant Intel in 2017, wrote that autonomous driving technology "will enable a new Passenger Economy worth $7 trillion in 2050." Just the consumer side of that business – so-called mobility-as-a-service, meaning robotaxis – will account for $3.7 trillion in annual revenue by 2050, the analysts wrote. It didn't take too long before the big automakers joined the party. General Motors , in a November 2017 presentation outlining the business case for its Cruise self-driving subsidiary, estimated that the total addressable market for robotaxis in the U.S. could hit $1.6 trillion in annual revenue once costs drop below $1.00 per mile, as it likely will once robotaxi services are widely available. (Ridesharing services with human drivers cost around $3.00 per mile, on average, the company estimated at the time.) The opportunity still looks huge in 2022 More than four years have passed since that GM presentation, but the company hasn't backed away from those trillion-dollar estimates. Cruise's CEO, Kyle Vogt, said at a Morgan Stanley conference in March that "we see a pretty easy path with our cost-down curve to $500 billion to $1 trillion market size – and that's the floor." It's the "floor," Vogt said, because that range assumes that consumers won't be moving away from private car ownership. What could trigger a mass move away from private car ownership? For Vogt and GM, it's all about cost. As the technology matures, as more autonomous taxis hit the road, the average cost per mile will come down. GM argued in 2017, and since, that the $1.00-per-mile point will likely be a tipping point for robotaxi adoption. But Vogt looks a little deeper, noting that the average cost of owning a car is likely between 60 cents and 80 cents per mile – and that car ownership comes with hassles like insurance and parking that disappear (for the consumer, at least) in a world where robotaxis are ubiquitous. "When you factor that in," Vogt said, "we think the TAM could be much, much bigger" than $1 trillion. 'Who will win' isn't the right question Some analysts still talk about a "race" to develop robotaxis and self-driving technology. But it has become clear that this won't be a winner-take-all space. The trillion-dollar-plus robotaxi pie will likely be split many ways. Bryan Salesky is a co-founder and the CEO of Argo AI, a Pittsburgh-based self-driving start-up backed by Ford and Volkswagen . Argo AI, Cruise, and Waymo — the Alphabet subsidiary created from the former Google Self-Driving Car Project — are viewed by most analysts as the companies most likely to deploy robotaxis at scale in the U.S. in the near future. Salesky has said the idea of a "race" to self-driving success is no longer useful — in part because many of the self-driving start-ups that survived the post-hype consolidation are aiming at different markets. "It's clear to me that different players are focusing on different miles, different use cases," Salesky told CNBC in a March interview. For instance, he said, "some have decided to focus on large over-the-road trucking. With the driver shortage and labor shortage, with the simpler highway miles, that's a big opportunity." But it's an opportunity for companies other than Argo AI, which is focused on going from address to address in an urban environment — a much more challenging environment for any driver, whether human or otherwise. Of course, companies like Waymo and Cruise are focused on the robotaxi space, too. But the idea that one (and only one) of these companies will "win" the robotaxi space just doesn't stand up to close scrutiny. To understand why, look at how ride-hailing has played out in New York City, where companies like Uber and Lyft compete with the city's trademark yellow cabs in a heavily regulated environment. Simply put, regulators and consumers haven't allowed any one company to take over the market – and that's not likely to change when the taxis' human drivers are replaced by algorithms. Why is it taking so long? Given the sheer size of the potential market, it's not surprising that automakers, venture capitalists and others have plowed billions of dollars into efforts to develop self-driving vehicles that can handle an urban environment safely. But it's 2022 and so far we've seen only a few limited deployments: What's holding them back? It isn't the cost of the hardware. One factor in the initial self-driving hype was the realization that key hardware components — small and inexpensive cameras and radar sensors — were already available, thanks to the huge global demand for things like smartphones. That 2013 Morgan Stanley report did note that lidar sensors were probably too expensive for mass deployment — and they were at the time — but low-cost automotive-grade lidar sensors are now coming to market. Prices have fallen from tens of thousands of dollars to around $1,000 in some cases. There are also upgraded wiring harnesses and other hardware components that are needed to make robotaxis at scale. While the cost of the on-board computing power needed to make a robotaxi work is still somewhat of a consideration, the real challenge has been creating the software to make it all work. As any engineer in the space will tell you, it's one thing to make a car drive itself around a quiet neighborhood, another thing entirely to be able to deal with all the uncertainties that arise in, say, rush hour traffic in downtown San Francisco. And it's yet another leap to be sure that your self-driving vehicle will handle that traffic more safely than an average human driver, day after day — well enough to convince the public to give it a try. Consider where things stood just three years ago, in early 2019. Back then, analysts expected companies like Waymo and Cruise to begin deploying robotaxis within months. Audi expected to launch a consumer vehicle with self-driving capabilities within a year or two. And Ford had plans to begin mass production of a self-driving commercial vehicle in 2021. As we now know, all those efforts (and many more) were delayed. While Covid may have been a factor to some extent, the real issue is that getting the software right has turned out to be a hard, hard problem. As former Cruise CEO Dan Ammann said in 2020, "Just getting to the minimum viable product, that initial vehicle that can drive more safely than a human, is probably the engineering challenge of our generation." It's likely to be at least a few more years before robotaxis become commonplace in most major U.S. cities. But that said, the leading companies in the space are making progress — and in a few areas, driverless taxis are already operating, though generally within strict limits imposed by cautious local authorities. For instance, Cruise is currently operating during late-night hours in parts of San Francisco while it works to secure permits for round-the-clock service throughout the city — a goal it expects to achieve before the end of 2022, GM CEO Mary Barra said during GM's earnings call on Tuesday. And Chinese search-engine giant Baidu just this past week became the first to win Chinese government approval to deploy its robotaxis , developed with Toyota -backed start-up Pony.ai, in parts of Beijing. Research spawned other innovations While the robotaxi "race" is still ongoing, it's worth noting that the massive R & D resources that have gone into robotaxi development over the last decade have already yielded technologies that work just fine in applications that are less demanding than urban taxi duty. "There have been ten years of immense investment and hype cycle around robotaxis," said Angus Pacala, CEO of lidar maker Ouster . "That investment has educated a generation of engineers that have now gone and found the less hyped, but easier to implement, applications across all kinds of automation." Specifically, Pacala said, the technology is finding applications where vehicles are operating in simpler environments — think mining vehicles, or local delivery robots, or small vehicles moving items around a big warehouse. The software required isn't nearly as complex as that for a robotaxi, but they're all autonomous vehicles — and as Pacala pointed out, they draw on many things that were initially developed or refined for robotaxis. "All of them use lidar sensors, cameras, radars, GPU compute and machine learning, and really similar algorithms," Pacala said. "It's all really based on — and comes out of — the mobility space, out of robotaxis." Investing in autonomous driving in 2022 What should investors make of all of this? While robotaxis didn't arrive as quickly as many people thought they would a decade ago, they are coming. Admittedly, it's still not clear exactly when they'll be widely available. And it's definitely not clear when — or whether — robotaxis will be so ubiquitous that consumers outside of cities will choose to give up their privately owned cars. More on the self-driving leaders Waymo, the company formed from the Google Self-Driving Car Project, is still a subsidiary of Google parent Alphabet. As of its last funding round, in June 2021, it was valued at about $30 billion. Cruise is a subsidiary of General Motors, which owns about 80% of the San Francisco-based startup. ( Honda Motor also owns a small stake.) Cruise was valued at about $21 billion as of March. Ford and Volkswagen hold equal minority stakes in Argo AI, which was valued at approximately $12.4 billion as of July 2021. Zoox is entirely owned by Amazon , which purchased it for about $1.2 billion in 2020. But for investors looking to hitch a ride, there are some clear opportunities. Aside from Tesla, the likely robotaxi leaders Waymo, Cruise, and Argo AI aren't public — at least not yet — the self-driving truck start-ups, automotive lidar makers, and other closely-adjacent companies included in the lists below could offer good opportunities to profit from the rise of autonomous vehicles.

A concept illustration showing the cockpit of futuristic autonomous car.

Metamorworks | Istock | Getty Images

It wasn't supposed to be this hard.

Morgan Stanley, in a lengthy 2013 report, told investors that self-driving taxis are "the new auto industry paradigm" and assured readers that "completely autonomous cars are set to available before the end of the decade."

But the decade has ended and robotaxis aren't a common sight on American streets. Amid the hype, it's easy to miss that the billions that poured into the sector haven't been wasted. Investors are already reaping the rewards of these efforts.