With Cruise In Rear-View Mirror, Robotaxis Power On Across The Globe

GM takes the game a step further, though, holding a new vision in which personally-owned cars can fully drive themselves – just like robotaxis do. This L4 capability is the new Holy Grail and whichever car company gets there first will have a huge market advantage. Such a promise has kept Tesla diehard customers in thrall for years (even while their patience is bewildering to the rest of us).

The Thinning Herd

In the U.S., four robotaxi companies have been leading the way: Cruise, Motional, Waymo, and Zoox. (While Tesla is worth watching in this space, they have not yet put forward details on their business case and on-road testing is at a very early stage.)

Waymo is the star player, operating a fleet of over 700 vehicles while providing 150,000 rides per week in Austin, Phoenix, Los Angeles, and the San Francisco Bay area. During 2024 alone, their riders have traveled over four million miles. I estimate that at least five million residents can access a Waymo robotaxi ride from their homes: about one percent of the U.S. population. Waymo is a business unit of Alphabet.

Waymo just announced it is making moves overseas. Their vehicles will arrive in Tokyo in early 2025. The aim is to gather mapping data and test their robot driver in a totally new environment: left-hand driving.

But can their vehicle drive safely? Waymo worked with Swiss Re to analyze liability claims related to collisions from 25.3 million fully autonomous miles driven by the company. The study used auto liability claims aggregate statistics as a proxy for at-fault collisions. Based on the results, Waymo asserts that the Waymo Driver significantly outperforms both the overall driving population and the latest generation of human-driven vehicles equipped with advanced driver assistance systems. Very promising.

Motional, formed in 2020 as a partnership of Hyundai and Aptiv, is regrouping after Aptiv pulled out and Hyundai took up the slack – to the tune of $475 million. The company hasn’t announced a date for initiating robotaxi operations in Las Vegas where they have been testing for several years. In a May 2024 blog post, President and CEO Karl Iagnemma said “we’ve updated our strategic plan to focus resources on the continued development and generalization of our core driverless technology, while de-emphasizing near-term commercial deployments and ancillary activities.” In September, Iagnemma stepped down, transitioning to the role of Senior Strategy Advisor. Chief Technology Officer Laura Major expanded her leadership role to become interim CEO and CTO. Nothing but question marks for these guys, at least for now.

Zoox, founded as a startup ten years ago, was acquired by Amazon in 2020. After launching safety driver operations in San Francisco, Las Vegas, and Austin, the company is now in the early stages of deployment in Silicon Valley. No information is available on their website about the size of their fleet or driverless miles thus far.

GM has a very long history with autonomy going back to the 1990’s when they led the National Automated Highway Systems Consortium, a cost-shared effort with the Federal Highway Administration. Subsequently, they funded a major effort with Carnegie Mellon University to compete in the 2007 DARPA Urban Challenge, in which they were the winner. In 2016, GM purchased Cruise for $1billion in 2016, with great fanfare. Cruise made steady progress, such that by 2023 the company was operating a substantial fleet of vehicles, racking up five million driverless miles in Houston, Dallas, Phoenix, and the Bay area. When a Cruise vehicle was involved in an unusual crash scenario late in the year, which was followed by a lack of transparency in reporting to California officials, a major reset was mandated by the mother ship. A full fleet shutdown and management shuffle ensued. A new CEO was hired and automotive veteran Steve Kenner was named as (their first) Chief Safety Officer. As Steve reported at a summer conference, their driverless-intent cars were dusted off and put into service with safety drivers starting last May. The company had their sights on re-starting fully driverless robotaxi services in 2025.

There is no indication that the final decision to close Cruise related to tech capability. But as the years went by after the acquisition, other views came to the fore in GM’s C-Suite. The shut-down was a hard-nosed business decision in which GM didn’t want to keep pouring cash into this endeavor and wait the many years it would take for Cruise to become profitable and competitive, when alternatives with more promising returns beckoned.

With Cruise’s exit, we now have one dominant player – Waymo – which is steadily growing their fleet to offer services in new cities. With Zoox in beginner-mode and the current uncertainty around Motional, Waymo is the only game in town for now.

But the robotaxi action in the U.S. is only a part of the picture. A vibrant robotaxi scene is roaring ahead in China, consisting of multiple well-funded players.

Robotaxis in China: Plenty of Buzz

In June 2024 the Chinese government approved nine automakers to test their systems in major cities. The China Society of Automotive Engineers foresees that one in five cars in China will have Level 4 autonomous capability by the decade’s end.

Over the years I’ve done my best to understand and report on the autonomy situation in the People’s Republic of China (PRC), but I’ve always felt that my efforts just scratched the surface.

Recently, I’ve had fascinating discussions with China tech-analyst Matt Johnson. Fluent in Chinese, he can scour the Chinese-language web to get insights and identify nuances that may not be apparent to the rest of us.

The leading robotaxi players in China are Baidu’s Apollo Go, Pony AI, and WeRide, with a newer player, Ruqi Mobility, aiming to break into the top ranks.

For this article, Matt provided a superb analysis of these companies, along with insights regarding the structure of underlying partnerships and how they contrast with the U.S.

Baidu’s fully driverless robotaxi business (Apollo Go, known as “Carrot Run” in China) is now operating in 11 cities, including Beijing, Shenzhen, Chongqing, and Wuhan. Apollo Go has driven over 28 million driverless miles and reportedly provided 8 million rides since the service’s inception.

Apollo Go provided approximately 988,000 rides in 3Q24, a 20 percent increase compared to the same period in 2023. In August 2023, the first robotaxi service between an urban area and an airport was launched by Baidu for Wuhan Tianhe International Airport. The Wuhan route was also the first instance of Chinese robotaxis providing trips using both urban roads and highways. In February 2024, Apollo Go received permission to operate on highways leading to Bejing’s Daxing International Airport.

Recently, there have been some misleading articles in the English-language press stating the Baidu will break even next year. Thanks to Matt, a closer look on the Chinese web provides clarity: “According to the head of Baidu’s autonomous driving unit, Chen Zhuo, Apollo Go is only expected to break even in the city of Wuhan, where Baidu first began to deploy its fleet,” Matt reports.

Still, Baidu leads in driving down the cost of robotaxi journeys. A 10-kilometer (6.2-mile) ride in a robotaxi in Wuhan costs between 4 and 16 yuan ($0.60 to $2.30), whereas an equivalent ride in a car driven by a human costs between 18 and 30 yuan ($2.47 to $4.11). Baidu’s sixth-generation robotaxi, the RT6, is priced below $30,000 per unit due to its modular design.

Based on Matt’s research, cost efficiencies like these noted above are expected to drive Baidu’s rapid robotaxi scaling in 2025. Wuhan, the capital of China’s Hubei province, is striving to become the world’s “robotaxi capital” as Apollo Go’s presence expands. In November 2024, Baidu subsidiary Apollo International was granted Hong Kong’s first autonomous vehicle testing license, allowing the company to conduct trials in the North Lantau area over a five-year period. Baidu has also signaled plans to expand Apollo Go into Singapore and the Middle East.

Matt’s take on Baidu: “Baidu Apollo is poised to lead China’s robotaxi market, leveraging its AI expertise, strong partnerships, and early-mover advantages. However, its growth could be limited by two defining characteristics of China’s tech landscape: intense competition and regulatory uncertainty.”

Founded in 2016 in Silicon Valley, Pony AI is active in robotaxi, robotruck, and personally owned vehicles. Pony.ai has partnerships with Chinese state-owned auto group FAW, Guangzhou Automobile Group (GAC), SAIC Motor, SANY Heavy Truck, and Sinotrans.

In November 2024, Pony AI listed on NASDAQ. The quiet period for stock analysts expired last week. According to Seeking Alpha, analysts including Goldman Sachs and Bank of America are now rating the stock as a buy.

Pony AI’s robotaxis have driven 2.4 million autonomous miles across urban roads and highways, and has obtained licenses to operate its driverless vehicles in the Tier-1 cities of Beijing, Guangzhou, Shanghai and Shenzhen. Prices for its services are reportedly comparable with traditional ride-hailing platforms, but PRC analysts have noted that they remain higher per customer unit than those charged by competitor Baidu. Pony AI is projecting to grow their robotaxi fleet in China from 250 vehicles to over 1,000. As part of the growth, Pony.ai is aiming to lower production costs of their robotaxis to less than $41,000.

In August 2023, Pony AI formed a joint venture with Toyota Motor (China) Investment Co., Ltd. (TMCI) and GAC Toyota Motor Co., Ltd. (GTMC) to mass produce robotaxis to support large-scale deployment. GTMC will provide Pony.ai with Toyota-branded battery electric vehicles, which are equipped with Toyota’s vehicle redundant systems that are suitable for L4 autonomous driving. The vehicles will be equipped with Pony AI’s advanced autonomous driving technology and will run on Pony AI’s robotaxi network platform, PonyPilot+.

In October 2023, Saudi Arabia’s NEOM regional development group invested $100M in Pony.ai to deploy robotaxis in NEOM’s regional projects and across the MENA region. To reduce exposure to U.S.-China tech competition, Pony AI has also announced plans to establish commercial operations in markets outside the United States, such as Singapore and South Korea.

Pony AI’s vehicle fleet consists of more than 150 robotaxis and 190 robotrucks. Launched with Sinotrans in 2021 as part of the Cyantron joint venture, robotruck operations are still the primary source of Pony AI’s revenue. Pony AI’s robo-trucks have accumulated 3.1 million miles to-date (with safety drivers on board). The company’s third source of revenue is its technology licensing and applications business.

Matt’s assessment of Pony AI is as follows. “A leading player in the development of L4 autonomous driving technology, Pony AI is also one of the largest operators. But the company’s business model is still in flux, with scaling and commercialization of its robotaxi business still in the early stages compared with its more proven robo-truck fleet and virtual driver technology.”

WeRide was founded in China in 2017. The company is backed by the Renault-Nissan-Mitsubishi Alliance, GAC Group, and Yutong Group, among others. Like Pony.ai, WeRide’s founding team includes alums of Tsinghua University, China’s preeminent research institution for autonomous driving systems. Significantly, their founders were part of the Baidu team before starting WeRide.

WeRide joined the NASDAQ stock exchange in September 2024.

WeRide operates over 600 robotaxis and self-driving minibuses, which operate in 26 cities worldwide. In November 2023, WeRide was approved to launch a paid service of fully driverless robotaxis in Beijing. The company is also active in Shenzhen and Suzhou.

WeRide’s global strategy underpins the company’s approach to growth. It was awarded a permit to test its AVs on public roads in the United Arab Emirates, with a safety driver, in July 2023. The UAE aims to make 25 percent of its transport autonomous by 2030. In late 2023, Singapore granted WeRide permits for testing its autonomous vehicles on public roads, also with a safety driver. In parallel, WeRide signed strategic cooperation agreements with Singapore’s Woodlands Transport Services and EZ Buzz, which offers bus services.

As CEO Tony Han told an interviewer in December 2023: “We are looking at developed countries with advanced information and transportation infrastructure, such as European nations, Japan, South Korea, and potentially lucrative markets in South America such as Argentina and Brazil. We’ve already started in the Middle East and are considering North African markets, including Morocco, as it is close to the Middle East.”

WeRide claims 1,200 days without any proactive safety incidents, and an accident-free track record across all of its global operations.

As Matt sees it, “WeRide’s IPO has positioned the company well for years of intense competition ahead, with its Baidu pedigree making it one of the most attractive China robotaxi companies to investors. Like competitors Baidu Apollo and Pony AI, WeRide is still searching for a break-even point that will turn revenue into profitability. With OEMs increasing their own investment in R&D, the pressure for autonomous systems companies to become commercially viable quickly is massive.”

The Iron Triangle: OEMs Key to Robotaxi’s Future in China

So, what about the role of the auto manufacturers when it comes to autonomy services in China?

Matt notes that “A quick scan of the most promising companies in China’s robotaxi ecosystem suggests that AV systems developers will end up dominating the space. Diving deeper, however, reveals that the success of these tech start-ups has been dependent on their partnership with legacy OEMs.”

In other words, Cruise’s collapse does not necessarily foreshadow the passing of traditional OEMs from the scene in China, especially given the power that these groups – many of whom are government-owned – wield in China’s state-managed economy.

Matt’s research shows that in 2019, WeRide introduced the term “iron triangle” to refer to the company’s vision of robotaxi development as an alliance between autonomous technology companies, OEM, and mobility platforms. Other robotaxi ventures in China which have followed WeRide’s iron triangle pathway include Yuanrong Qixing/Dongfeng Motor/Caocao Travel, and Momenta/SAIC/Xiangdao Travel.

WeRide has introduced variations on this theme as well. For example, combining its technology with existing taxi fleets, as it did in Guangzhou with Baiyun Taxi Group in 2019.

Though still in use, the “iron triangle” model may already be less ubiquitous than it was several years ago. OEMs, however, appear just as prevalent. Some contemporary PRC analysts now divide the autonomous driving space into three distinct types of companies: tech start-ups, OEMs, and legacy tech giants (e.g. Alibaba, Baidu, and Huawei). They also note that there are specific weaknesses in the “iron triangle” model, specifically concerning who pays for the costs of commercialization until break-even is reached.

“In this environment, OEMs have the comparative advantages of being able to both manufacture their own vehicles and having relatively deep pockets. Despite Cruise’s exit, the example of China’s robotaxi partnerships suggests that vehicle OEMs will remain crucial partners for PRC-based tech companies seeking to commercialize and expand,” Matt said.

How was the Cruise shutdown viewed in China?

Matt provides the fascinating insight that, “rather than rejoicing in the demise of a U.S. company, PRC auto industry analysts/observers on social media seem to be stepping back and reflecting on Cruise’s demise as evidence of challenges in AV development generally. That being said, there is definitely a sense that legacy OEMs like GM do not seem as credible in this space as upstart peers … or China’s own new breed of champions.” He adds that there is significant skepticism around whether GM’s “pivot” to an advanced Level 3 Super Cruise is credible; or is it simply a “face-saving” move?

Finally, questions still remain around what technical pathways are most likely to lead to fully autonomous driving at scale. China Unicom and Huawei have recently launched China’s first 5G-A V2X innovation base to pilot hybrid “vehicle-road-cloud integration” solutions as the future of national transportation networks. This approach combines ultra-fast 5G-A networks, intelligent infrastructure, and cloud computing to enable seamless communication between vehicles and their environment. Similar activities are occurring at small scale the U.S. and Europe. While it’s too early to judge the results of initial deployments, the topic bears watching over the next years.

Where Are We Now?

In GM’s announcement of the Cruise shutdown, CEO Mary Barra noted that the robotaxi market is becoming “increasingly competitive.” Looking through a global lens, the above discussion illuminates the picture.

Is robotaxi in trouble? That’s not how I see it. If not for the Chinese market, I’d be worried.

And based on Matt’s reporting, the Chinese autonomy companies will expand to any market that welcomes them. The result: robust availability of autonomous trips in developed nations is very much in the cards during the remainer of the 2020’s.

Nevertheless, the market will remain “warped” in the U.S. due to concerns about Chinese tech companies spying. The U.S. government wants no part of roving tech devices gathering data on U.S. cities. This leaves us with the in-country players. Having a single dominant player in any tech space is not ideal. Yet, Waymo’s performance and growth is fascinating to watch. In San Francisco, their market share is now equal to Lyft. Incredible.

Cruise could have re-started operations and built a substantial fleet, but are now out of the picture. While Waymo powers ahead, Zoox is moving at a conservative pace. The future of Motional is unclear.

How does the “iron triangle” paradigm play in the U.S.? Zoox is owned by tech-focused retailer Amazon. Waymo is owned by tech megalith Alphabet. Only Motional is owned by a car-maker, Hyundai.

Not there there aren’t other car OEMs in the mix. Volkswagen, Nissan, and others have shared plans for robo-shuttle deployments in their home markets.

What is “Over”?

What’s “over” is not robotaxi. This is a market rationalization, reshuffling what type of company, in what business culture, is fit to stand up and operate a robotaxi fleet.

Importantly, we mustn’t forget a new breed of autonomy actors: those who develop the tech and license it to fleet operators or vehicle manufacturers. In recent years, this space has become super-active with well-funded companies: Nuro, Wayve, and Turing are prime examples. As Nuro COO Andrew Chapin put it in a recent On The Road to Autonomy podcast, “The adaptable Nuro Driver targets partnerships with mobility platforms and fleet operators, positioning Nuro to power robotaxis, delivery services, and more in the coming years.”

The Cruise shut down, rather than hand wringing and gnashing of teeth, has instead been more like a shrug. We’re moving on.

Where Are We Going?

In 2025, Waymo will expand to new cities and also become active in Japan. In parallel, the evolution of Motional, and Hyundai’s appetite to pour in the funding to stand up robotaxi services will be fascinating to watch.

Initiation of robotaxi services in the UAE is highly likely next year and services within China will expand.

Commenting further on Cruise, I must say that I’m quite sad to see this turn of events. Cruise’s self-driving tech was very good and getting better post-reset. All the capabilities were in place to put forth a solid robotaxi offering. It’s going to take significant time and money for GM to release a Level 4 product, if they ever do so.

Car companies exist to make and sell cars. There is no fundamental forcing function to offer autonomy. They will offer new features to delight the customer and maintain or grow market share. As they develop these features, they must evaluate the cost to develop the systems as well as the cost to the customer, all while assessing risk. What public safety and liability nightmares will ensue when GM, Tesla, or another automaker sell fully autonomous driving to anyone that wants it?

Alphabet, Baidu, and the other tech companies exist to make tech breakthroughs and establish new markets and services over a long timespan. Rather than selling vehicles to consumers, they operate fleets to provide mobility.