It has been 100 years since Francis Houdina operated his radio-controlled automobile through New York, barreling down Broadway and Fifth Streets.
While the vehicle – claimed by some to be the first “driverless” car – narrowly missed several accidents, it finally ended its history-making run by crashing into another car.[1]
In the subsequent decades, many attempts at autonomous and semi-autonomous driving technology arose, but widespread interest did not take hold until the 1980s when the US Department of Defense became involved via its Defense Advanced Research Projects Agency. The 2004 DARPA Grand Challenge tasked engineers, students, and more to create an autonomous vehicle to navigate a desert race course. Though no vehicle finished the entire course, the competition validated the concept of autonomous vehicles, and companies began setting their sights on leading the charge to market. Since then, significant milestones have contributed to the evolution of autonomous driving, including Google’s Waymo’s 2015 achievement of the first successful fully autonomous trip on public roads.
More than US$100 billion has already been spent on autonomous driving, the majority of which is in the robotaxi sector.[2] However, after five years of service, operators of Level 4 robotaxis—autonomous vehicles that can drive without human intervention in specific conditions—have yet to see a profit. This article examines the dream of robotaxis and what it will take to move them from hype to reality.
Technologies Behind the Vehicles
According to research firm IDTechEx, the global robotaxi vehicle market value will reach US$174 billion in 2045 and show a 20-year compound annual growth rate of 37 percent between 2025 and 2045.[3] The US and China are expected to dominate market share. Such rapid growth requires many technologies, including radar, lidar, and ultrasonic sensors and cameras, to provide such visual data as object detection and identification of vehicles, pedestrians, and obstacles. Processors, connectivity protocols, and actuators also play important roles; more recently, artificial intelligence and machine learning enabled vehicles that tap into their experiences and complex algorithms to make sense of the volumes of data gathered.
AI and ML also improved accuracy and reliability by analyzing data in real time, recognizing potential hazards, and predicting the behavior of other drivers, all while determining the best course of action. Vehicles make informed decisions, navigate extremely tricky environments, and become increasingly proficient at navigating scenarios. This is in part thanks to global positioning system receivers that access signals from multiple satellites to triangulate position. These data, combined with a car’s speed and direction, document exact locations in real time.
Using high-precision mapping systems, companies can provide road layouts, lane markings, traffic signs, and speed limits. Those data allow the cars to anticipate upcoming turns, intersections, and other potential obstacles to plan movements in advance. Robust software systems are implemented to detect and prevent such risks and malfunctions as errors in sensor readings and cybersecurity threats, playing a crucial role in maintaining safety.
Behind these technologies are several large companies working on autonomous vehicle technology individually and in partnership. Market leaders Waymo and Apollo have a strong reputation for their sensors, maps, and intelligence. Waymo and Baidu use complex architectures that combine several AI systems for decision-making. In comparison, Tesla uses a camera-based approach supplemented by maps and an end-to-end deep learning system that processes raw sensor data to make driving decisions. This approach has been controversial, with some experts saying it limits Tesla’s ability to achieve the precision levels seen by its competitors.
As autonomous vehicles take to the road, commercial driverless robotaxi services are gaining momentum in the US and China. Key players in these regions include Google’s Waymo, Baidu’s Apollo Go, and Amazon-backed Zoox, with the number of providers and locations increasing annually.
Roadblocks to Adoption
While technological precision potentially translates to lower accident rates than human-driven cars,[4] challenges exist. Regulatory and legal challenges, safety, data privacy, accident liability, and system certification are necessary, expensive, and inconsistent regionally. Additionally, safety and reliability requirements in urban settings mandate that vehicles be able to handle bad weather, complex traffic conditions, and human behavior.
For example, General Motors subsidiary Cruise paused its entire robotaxi operation shortly after an incident involving a pedestrian in 2023.[5] This event prompted debate about how ready we are for autonomous vehicles and emphasized the delicate relationship between public perception, safety, and regulation.
These challenges continue to delay robotaxi market penetration and the establishment of a viable roadmap for a profitable sector. Governments and the private sector will need to continue to support the significant financial investment necessary for long-term solutions.
Missing Profits
Using autonomous vehicles as robotaxis remains unprofitable. Investment and revenues will need to provide gross margins sufficient to cover research, development, and overhead costs, as well as the largest expense—personnel. The Cruise business model required 1.5 personnel members per vehicle to assist in driverless operations.[6] These personnel included call center staff, teleoperations personnel to provide remote guidance to autonomous vehicles, roadside support to recover vehicles, and operations staff at depots to service and charge vehicles.
Based on the costs of that model, companies operating with just one person per vehicle will see an overall loss of $34,000 per autonomous vehicle.[7] Autonomous performance continues to improve, but profitability will drive autonomous vehicles to wide-scale deployment. The current lack of robotaxi profitability will inhibit larger fleets, making it difficult to reduce costs and subsequently increase widespread adoption. However, Baidu claims its Apollo will likely reach profitability in 2025, before its US counterparts.[8]
Tesla Enters the Robotaxi Market
Tesla is busy racking up miles to prove its robotaxi approach. By April 2024, it had more than 1.3 billion miles of experience under full self-driving mode, adding one billion miles every two months.[9] To date, Tesla’s FSD mode is available in the US and Canada and is expected to roll out in China in 2025.
In October 2024, the company unveiled its long-awaited robotaxi, Cybercab, and predicted it would be in production before 2027 at a price tag of less than $30,000.[10] The vehicles rely on cameras that are less expensive than radar and lidar sensors, and AI—trained by the raw data it collects from its millions of vehicles—will teach its cars to drive. Tesla also unveiled another prototype for a Robovan for up to 20 passengers in the future. However, shortly after the Cybercab announcement, the US National Highway Transportation Safety Administration announced an inquiry into Tesla’s FSD after the company reported four collisions in low-visibility conditions.[11]
Market Readiness
The actions of the top three players and their approaches and successes to date will help determine how ready the market is for robotaxis. Waymo has a head start, huge investments, patents, and millions of logged autonomous miles. It excels in mapping, data analysis, testing, and deployment. Challenges include sensor fusion, dealing with unexpected scenarios, and hardware reliability. Its partnerships include Zeekr for electric vehicle development and collaboration with Uber to expand market reach. Chinese tech giant Baidu operates its Apollo Go robotaxi service in multiple cities across China. Strengths include cutting-edge AI and mapping capabilities; however, challenges are seen in complex traffic conditions and strict regulations. Tesla focuses on enhancing its Autopilot and FSD capabilities for fully autonomous robotaxis and transitioning to sustainable energy. With its existing fleet, however, Tesla faces challenges in perfecting FSD technology, including navigating complex traffic scenarios, adverse weather, unexpected obstacles, utilization, charging infrastructure, maintenance, the political landscape, and its tendency to overpromise and under-deliver.
Public readiness is influenced by safety concerns, accessibility, and the education needed to demystify the technology. A recent survey by the American Automobile Association found that 68 percent of Americans are hesitant to ride in a fully autonomous vehicle due to trust issues.[12] Companies like Waymo are attempting to address this by offering free trials in specific cities, while Tesla leverages its vast fleet to gather data and improve perception over time.
Regulatory hurdles, ethical considerations, and shaky public acceptance must be solved before robotaxis find significant adoption. Safety standards for autonomous vehicles must match or exceed those established for human-driven vehicles. Accident liability will need to be addressed on several fronts, including robotaxis, manufacturers, and operators, while insurance policies will need to cover technology failures and even cyber threats. To provide adequate infrastructure, planners will need to ensure reliable 5G and vehicle-to-everything networks for real-time data transfer, account for changes to traffic flow, establish pick-up and drop-off zones, and provide adequate charging support.
Conclusion
The future of robotaxis remains a complex challenge. As Waymo, Tesla, and Baidu ramp up production and testing, many obstacles will prevent them from becoming profitable and widely adopted. How quickly these changes happen will depend on public confidence and readiness, regulatory compliance, and sustainable scaling. As the industry grows, only a handful of dominant companies will lead the next generation of autonomous transportation. No one will know whether $100 billion in investments has paid off until necessary advances are made, but it will certainly open the door for innovations that redefine transportation.
By Carolyn Mathas for Mouser Electronics – Published February 17, 2025
Sources
[1]https://www.discovermagazine.com/technology/the-driverless-car-era-began-more-than-90-years-ago
[2]https://www.idtechex.com/en/research-report/autonomous-vehicles-markets-2025-2045/1045
[3]https://www.idtechex.com/en/research-report/autonomous-vehicles-markets-2025-2045/1045
[4]https://www.frontiersin.org/journals/built-environment/articles/10.3389/fbuil.2024.1383144/full
[5]https://www.businessinsider.com/robotaxis-general-motors-cruise-problems-tesla-elon-musk-2024-12
[6]https://www.nytimes.com/2023/11/03/technology/cruise-general-motors-self-driving-cars.html
[7]https://www.forbes.com/sites/gustavo-castillo/2024/10/09/challenging-economics-will-slow-the-deployment-of-robotaxis/
[8]https://www.theregister.com/2024/05/17/apollo_go_profitable/
[9]https://www.teslarati.com/tesla-fsd-users-pass-1-3-billion-cumulative-miles/
[10]https://www.marketwatch.com/story/tesla-robotaxi-day-is-here-5-things-to-watch-for-at-the-we-robot-event-0ecbfa42
[11]https://www.reuters.com/business/autos-transportation/nhtsa-opens-probe-into-24-mln-tesla-vehicles-over-full-self-driving-collisions-2024-10-18/
[12]https://newsroom.aaa.com/2023/03/aaa-fear-of-self-driving-cars-on-the-rise/
