At a certain stage of autonomous driving development, a viewpoint emerged suggesting that dominance in the automotive industry might shift from automakers to software companies. However, interestingly, as Robotaxis approach large-scale deployment, autonomous driving companies find themselves increasingly reliant on traditional automakers.
On October 30, Waymo and Toyota announced they are advancing their autonomous driving collaboration. The two parties plan to launch testing and initial operations of an autonomous fleet based on the Toyota bZ4X in Japan starting in 2027. Unlike previous modifications made to mass-produced vehicles, this batch will be manufactured directly at Toyota's Takaoka Plant, with Waymo's autonomous hardware also integrated at the same facility.
Behind this event lies a question worth significant attention.
Previously, the autonomous driving industry was most concerned with who could enable vehicles to drive without human intervention. As Robotaxis move from technical validation toward larger-scale deployment, the focus of competition shifts to who can truly turn autonomous driving capabilities into a continuously replicable product.
This clearly depends not only on algorithms.
01. Why Software Cannot Exist Without the Vehicle
The concept of "Software Defined Vehicles" (SDV) has been proposed for quite some time. Many people perceive this concept as automakers providing hardware platforms while technology companies handle software; as software capabilities steadily improve, the importance of the vehicle itself would gradually decline.
However, the development of autonomous driving does not seem to have followed this path entirely.
The performance of an autonomous driving system is difficult to discuss independently of the specific vehicle. Steering and braking characteristics influence control strategies, body structure determines sensor mounting space, and power supply and thermal management capabilities relate to whether the computing platform can operate stably over the long term. Therefore, even if the same autonomous driving solution is applied to different models, it still requires re-adaptation.
This relationship becomes even tighter during the mass production phase. Prototype vehicles can quickly undergo modifications by adding sensors, computing units, and wiring harnesses, but this approach is difficult to replicate directly in large-scale production.
Where sensors are placed, how the computing platform dissipates heat, and how key actuation systems achieve safety redundancy all need to be considered early in the vehicle development stage.
This is why the more complex the autonomous driving capability, the higher the requirements imposed on the vehicle platform.
Software has not turned cars into standard hardware that can be arbitrarily swapped; instead, it has brought autonomous driving systems earlier into the vehicle development process.
Waymo and Toyota's decision to complete base vehicle production and autonomous hardware integration in the same factory indicates that autonomous driving hardware is moving further from post-production modification into the automotive manufacturing workflow.
Although this does not require every autonomous driving company to participate in vehicle manufacturing themselves, it indirectly confirms that the vehicle is no longer just a carrier for the autonomous system; it has become part of the autonomous driving product itself.
02. Scaling Up Is More Than Just Building More Cars
Enabling one car to drive autonomously and enabling ten thousand cars to operate stably long-term are two completely different concepts.
During the technical validation phase, the core task for autonomous driving companies is to prove whether the system can drive one car well. Even if issues arise, engineering teams can address them individually.
But after the fleet size expands, the nature of the problems changes.
If a sensor solution requires complex manual installation and debugging, costs will escalate rapidly as the number of vehicles increases; if a hardware component lacks reliability, the impact extends beyond a single test vehicle to potentially affect the operational efficiency of an entire batch.
At this point, what autonomous driving companies face is no longer just a technical problem, but a product problem.
Technical validation seeks to determine if this car can run; scaling up seeks to ensure every car can run with near-consistent quality and cost.
The difference between these two dictates why the role of traditional automakers is more irreplaceable than ever.
Vehicle development solves how to integrate complex systems into the product beforehand; scale manufacturing solves how to ensure consistency across a large volume of vehicles; supply chain and quality systems determine whether these vehicles can operate stably long-term.
For Robotaxis, these capabilities directly impact the business model.
A Robotaxi is not a one-time sale upon delivery, but a long-term service operation. Vehicle reliability, maintenance efficiency, and downtime eventually translate into operating costs.
Therefore, as autonomous driving enters the scaling phase, the competitive logic of the industry is also changing.
Algorithmic capability determines whether the vehicle can start driving, but product and manufacturing capabilities begin to determine whether the system can expand at a reasonable cost.
This is perhaps the most noteworthy aspect of the new Waymo-Toyota partnership.
The focus of the collaboration is no longer just providing a vehicle model suitable for autonomous driving, but attempting to incorporate the autonomous driving system directly into a mature manufacturing system.
03. Who Defines an Autonomous Vehicle?
Whether looking at Momenta's controlling stake in the SAIC Intelligent Driving joint venture discussed previously, or today's Waymo-Toyota collaboration, viewing these events within a broader industrial context reveals that the boundary of division between autonomous driving companies and automakers is shifting forward.
In the past, the relationship between automakers and autonomous driving companies could be simply understood as: automakers first develop a vehicle, and then autonomous driving companies install their systems onto it.
This model suits technical validation and early small-scale deployments.
But when the goal becomes large-scale operations, many issues cannot wait until vehicle production is completed to be resolved.
How sensors and computing platforms are integrated, how power supply and thermal management are designed, and how vehicle actuation systems meet autonomous driving needs—all these questions must be considered early in the product development stage.
Consequently, what an autonomous vehicle looks like is no longer determined solely by automakers, nor is it just a matter of autonomous driving companies providing a software suite.
Both parties are beginning to jointly participate in product definition. This may be the change occurring in the next stage of the autonomous driving industry.
In the past, people debated whether software would redefine cars, worrying that traditional automakers would ultimately degrade into mere hardware manufacturers.
But now, it appears that autonomous driving has not simply split the automotive industry into technology companies handling intelligence and automakers handling manufacturing.
The more autonomous driving moves toward scaling, the tighter the relationship between software and the whole vehicle becomes.
Technology companies need to understand vehicles earlier, and automakers need to understand autonomous driving earlier.
In the future, true competitiveness may not lie in who alone possesses the strongest algorithms or the largest manufacturing scale, but in who can integrate software capabilities, vehicle platforms, and manufacturing systems into a repeatedly replicable product earliest.
The new Waymo-Toyota partnership provides a window to observe this shift.
Waymo possesses mature autonomous driving technology and real-world operational experience, while Toyota holds vehicle development and large-scale manufacturing capabilities. What both parties now need to solve is no longer just getting autonomous driving technology onto cars, but how to bring this technology into a larger market.
Toyota and Waymo also stated that they will continue to explore applications of autonomous driving technology in Toyota's personal passenger vehicles in the future, and study how to utilize data and technical insights generated from Waymo's real-world operations to improve road safety. This suggests that the relationship between the two may not stop at supplying vehicles for Robotaxis.
For Waymo, Toyota provides not just a car, but an industrial capability to stably manufacture complex products. For Toyota, Waymo provides not just an autonomous driving system, but technical and data experiences accumulated through long-term real-world operations.
In the past, the industry was more concerned with whether autonomous driving would change the dominance of the automotive sector. Now, as autonomous driving truly moves toward scaling, it is becoming difficult for anyone to define an autonomous vehicle alone.