Waymo Than Meets the Eye
By: Jarry Wu
The Ivey Business Review is a student publication conceived, designed and managed by Honors Business Administration students at the Ivey Business School.
Taxi! Taxi! Robo Taxi?
In October 2020, Waymo became the first company to offer fully autonomous ride-hailing services to the public (Waymo One). Originating from Google’s self-driving car initiative in 2009, Waymo has become the industry leader in pioneering commercial autonomous vehicles (AV). This advantage stems from their incredible in-house talent, pricing advantages for cloud infrastructure as an Alphabet Inc. subsidiary, and abundant access to financing.
These advantages have allowed Waymo to invest upwards of $11.10 billion into improving the safety and reliability of the technology behind their autonomous driving system, the Waymo Driver (WD), while also focusing on optimizing its sensor suite to be more accurate and cost-effective. These innovations mean it costs Waymo ~$0.30 per mile to offer their services compared to Uber’s estimated $0.69 per mile and allows them to charge, on average, 6 percent–11.5 percent less per mile than Uber. This difference reflects one of Waymo’s most significant competitive advantages in ride-hailing: robots don’t need to be paid. Approximately 85 percent of Uber’s revenue per ride is given back to its drivers. Although Uber has included additional charges that appear to bring their take rate closer to 41 percent (compared to the official 15 percent), Waymo still has the clear pricing advantage of lower operating costs and the ability to capture nearly 100 percent of the revenue earned from serviced trips.
Waymo’s latest 6th gen Geely/Zeekr (G/Z) robotaxis are also expected to reduce costs significantly. Using Baidu’s Apollo RT6 as a proxy, the cost of the G/Z robotaxis falls below $28,000/vehicle, as the RT6 uses twice as many radar and LiDAR sensors as Waymo’s G/Z robotaxis. It is also important to note that this cost figure excludes potential economies of scale for high-cost hardware like LiDAR sensors, which are forecasted to drop from $5,000 to ~$200 with an increase in purchase volume, decreasing the hardware bill-of-materials below $2,000 per vehicle. Looking at Robotaxis’ operating cost advantages and potential economies of scale, it can disrupt the ride-hailing industry, and Waymo is in a strong position to lead this charge.
Rearview Mirror on the Wall: Who’s the Best Ride-Hail Service of Them All?
Despite Waymo’s operating cost advantages and opportunity to capitalize on economies of scale, being an autonomous robotaxi service is not the most viable long-term strategy for them because of one key fact: Waymo has no large-scale manufacturing experience. Scaling a fleet to the hundreds of thousands of vehicles necessary to service ride-hailing at a scale like Uber’s would require an additional ~$26.29 billion in capital expenditure, even with cheaper vehicle models. Fleet management, maintenance, ride-hailing infrastructure, and regulatory compliance would also represent high-cost logistical issues that could risk slowing down Waymo’s progress in autonomous driving AI and ML, which has been their core strength as a company for 15 years. For example, the maintenance costs alone of delivering an Uber-scaled fleet of AVs sits at ~$4.15 billion per year, which would force them into diverting a lot of resources into manufacturing and upkeep rather than self-driving technologies.
Waymo is at a crossroads. To continuously improve the reliability and performance of their AI driver, they need to scale their fleet size to collect exponentially more training data. However, without any manufacturing capabilities, they are forced to rely on existing automakers to build a fleet capable of collecting the data necessary to improve their WD. Based on Waymo’s projected on-road testing date for Hyundai’s IONIQ 5 AVs, it takes one year per manufacturing partnership for Waymo to integrate their self-driving technology with an automaker’s tech stack, and calibrate their sensors to the new models. Being forced to tailor their self-driving solution to meet the needs of different automakers throttles Waymo’s speed of innovation and gives time for competitors like Cruise to re-enter the market. The faster Waymo can incentivize automakers to adopt their WD technology, the more data they can collect on driving behaviour and city mapping, and the harder it becomes for competitors to build a comparable product.
The Hitchhiker’s Guide to Waymo Driving
Waymo’s dominant autonomous driving technology relies on two key pieces of software: the Waymo Driver (WD), an advanced system of algorithms and AI models that work together to facilitate autonomous driving, and Simulation City (SC), Waymo’s proprietary simulation software used to train the WD.
1. Double Down on Software
What sets Waymo’s software apart from competitors is the way WD and SC interact with each other. SC allows the WD to practice navigating rare and challenging driving scenarios, from something as small as raindrops on its sensor to as complex as afternoon solar glare. SC is also constantly refining its performance with real-world data being collected by Waymo’s existing AV fleet. Unlike competitors like Motional or Cruise, who rely on their AVs encountering unlikely scenarios in real life before refining the software, Waymo could proactively train WD on novel risky scenarios by creating them in SC.
2. Exit the Business of Ride-Hailing and enter the Business of Software Licensing
Considering the strength of Waymo’s software systems, it becomes clear that the solution lies not in trying to compete head-on against other companies interested in displacing Uber as the de facto ride-hailing service but instead in embracing their true strength: autonomous driving systems. In tandem with their fleet of AVs, Waymo should begin licensing access to its self-driving software solutions (i.e. WD and SC) to automakers like GM who are interested in entering the robotaxi industry instead of solely focusing on expanding their operations through relatively slow partnerships and larger fleet sizes.
3. Develop a Standardized Software Package for Integration with Automakers
One of the most significant challenges in AV development is the integration of self-driving software into a vehicle’s existing electrical control units (ECUs) and controller area network (CAN) buses. To address these issues, Waymo should also begin developing a software package that automakers could upload into their ECUs and CAN buses that would help standardize messaging for key parts of the vehicle that are integral to autonomous driving (i.e. acceleration, braking, steering, lane keeping, emergency responses). These two strategic pivots would allow Waymo to scale much faster than it currently is while collecting exponentially more data that could help improve the reliability and efficiency of WD, allowing it to solidify its place as the gold standard for autonomous driving systems.
To Adopt the Waymo Driver or Not to Adopt, That is the Question
As of 2024, GM has spent over $8 billion on autonomous driving, Toyota has invested $3.30 billion, and Ford has committed $7 billion towards self-driving technologies. Around the same time frame, GM’s Cruise cars secured a partnership with Uber, Toyota announced their e-Palette autonomous vehicle concept while emphasizing the CASE era of vehicles, and Ford established Latitude AI as a subsidiary focused on autonomous driving systems as well as Ford Smart Mobility LLC. This trend heavily implies that America’s three largest automakers are beginning to look for ways to enter the mobility-as-a-service (MaaS) industry through AV fleets and autonomous ride-hailing services as a means of diversifying revenue streams and increasing brand reputation.
Waymo is currently on track to spend $11.1 billion on its AV fleet and technology and is already capable of providing fully autonomous driving services in multiple cities. American car manufacturers are already behind in AV technology, and using Waymo as a proxy would require billions more investment before they reach commercial-grade autonomous driving like Waymo’s. The Waymo Driver is a more robust and reliable autonomous driving software than automakers have developed themselves. Therefore, the benefits of a software licensing business model are two-fold. It would allow Waymo to shed its dependence on individual auto manufacturers and allow automakers entering the MaaS industry to focus on areas where they hold competitive advantages, primarily in vehicle design, creation, and distribution.
Toyota’s e-Palette buses, for example, are built to be local mobility solutions from shuttle services to mobile storefronts. However, because Toyota has yet to develop the self-driving technology necessary for their e-Palettes to drive on their own, they are relegated to following a fixed route within their plants. With access to WD and SC, Toyota could realize the full potential of its e-Palette mobility solution much sooner and allocate its full manufacturing force behind the production of these vehicles. This would rapidly accelerate release timelines for the e-Palette and allow them to more quickly realize their goal of becoming a “mobility company.” Similarly, both Ford and GM have expressed interest in becoming mobility companies. This suggests that Waymo would only need one automaker to adopt their technology standards to incentivize competing companies to do the same, likely due to their fear of being left behind in the transition towards mobility-focused solutions beyond private cars.
With this business model, Waymo could also offload the capital needed for fleet expansion to several large-scale manufacturers who could harness their economies of scale to design, build, and distribute these vehicles faster and cheaper than Waymo could themselves. As these automakers begin entering the market with their fleet of AVs built on WD’s autonomous driving capabilities, Waymo would be gathering exponentially more data on real-miles driven that could be used to improve WD’s adaptability to different vehicles and SC’s ability to simulate several different cities and environments. This rapid expansion of their AI capabilities would provide Waymo with a powerful competitive moat that would be impossible for other AV software solutions to replicate without the vast datasets Waymo would have collected.
Houston, We Have Robots Driving Cars
To ensure Waymo’s ECU and CAN bus software packages are adopted by automakers to more easily integrate WD into their AV fleets, Waymo should sell these packages at or below cost as a loss leader. Using Rivian as a proxy for the number of ECUs needed for autonomous driving, Waymo uses an estimated seven advanced ECU computers to further the autonomous driving capabilities of their current vehicles. Although the ECU R&D expenses for Waymo are undisclosed, the cost to develop and integrate a package of advanced network-level ECUs and CAN buses is close to 35 percent of Waymo’s entire electronic system R&D efforts. Traditional original equipment manufacturers (OEMs) adopt a royalty-based pricing strategy, charging automakers for each car produced with their software. However, Waymo is not reliant on its ECU and CAN bus software packages to generate revenue. Therefore, it could offer a fixed fee of around 35 percent of the total R&D expenses spent on ECU and CAN bus integration efforts. This pricing strategy would allow automakers to spread the cost of adopting Waymo’s ECU package across their entire fleet, incentivizing WD over a more expensive and less capable in-house solution.
Waymo should open access to their primary revenue generator's WD and SC software via APIs and use a per-mile pricing model. Waymo’s funding recently totalled $11.10 billion, with miles driven (real world and simulated) adding up to around 20.02 billion miles, ~20 billion of which were driven in Simulation City. Assuming that 70 to 80 percent of the funding was spent developing the Waymo Driver and the other 20 percent to 30 percent spent on building, maintaining, and improving Simulation City, Waymo’s estimated cost per mile for developing the latest version of the Waymo Driver is around $0.39 per mile–$0.44 per mile. Consequently, the costs for Simulation City would be an estimated $0.11 per mile–$0.17 per mile. With estimated profit margins of 15 percent, Waymo should be pricing their Waymo Driver technology at $0.46 per mile–$0.52 per mile driven and Simulation City access at $0.13 per mile–$0.20 per mile.
Waymo’s total miles driven as of 2024 are an estimated 20.02 billion miles. This means the Waymo Driver and Simulation City would generate $9.21 billion–$10.41 billion and $2.6 billion–$4 billion in revenue, respectively. Therefore, based on this business model, Waymo would have generated $11.81 billion–$14.41 billion over its 15 years of operations.
Drivers? Where We’re Going, We Don’t Need Drivers
Based on its existing operations, Waymo’s business model aims to disrupt the ride-hailing industry as it develops self-driving technology capable of displacing Uber-like services. This article highlights a more realistic trajectory for the company by focusing on its core competency in self-driving software solutions. Legal obstacles and concerns around public perception are also slowly fading as consumers grow more comfortable hailing AVs and regulators become more trusting of self-driving technology. So, don’t be surprised if the next time you hail a ride, you find yourself in a car built by Ford, driven by Waymo, and managed by Uber, because this future may just come much sooner than you might expect.
