As the development cycle for a new vehicle is compressed to 12 months, how much time is truly left for suppliers?
On September 17, at the 2026 AI-Defined Vehicle Forum, Wang Xianbin, Partner at Gasgoo and Vice President of the Gasgoo Research Institute, provided a figure: it may be only 3 to 4 months.
Over the past few years, the pace of new vehicle launches in China's automotive industry has been accelerating. The time required for a new vehicle from project initiation to market launch is being steadily compressed, shifting from the traditional three- to four-year development cycles of legacy automakers to 18 months, 15 months, or even 12 months.
However, a shorter timeline does not mean that the required design, development, and validation work is proportionally reduced.
Wang Xianbin stated it very directly, "If all these time savings still fall short of meeting the milestones, the only option left is to cut into the validation phase."
Compared to the question of "how much R&D efficiency AI has improved," this statement perhaps deserves more attention.
It highlights a more practical issue emerging as vehicle development speeds up in China: once all efficiency-boosting processes have been accelerated, where else can time be squeezed from?
Building a Car in 12 Months: What Is Ultimately Compressed?
In Wang Xianbin's view, the rapid shortening of the vehicle development cycle in China stems primarily from changes in the development mechanism.
Some emerging automakers have shorter decision-making chains, with core management directly involved in product definition—"the boss is essentially the product manager." Meanwhile, supplier involvement is occurring increasingly earlier in the process. Software, hardware, and full-vehicle systems have shifted from the predominantly sequential workflows of the past to parallel development.
AI is further integrated into this process.
From styling design and material R&D to crash simulation, software testing, virtual validation, and OTA iterations, an increasing number of R&D processes are leveraging AI and digital tools to boost efficiency.
Yet even so, time may still fall short.
Wang Xianbin noted that if a vehicle is developed within approximately 12 months, the actual time left for suppliers to complete development and validation may be only 3 to 4 months.
In vehicle R&D, certain phases inherently possess time rigidity.
Take winter calibration as an example. A prospectus disclosed by CATARC this year indicates that a complete winter calibration R&D validation for a vehicle model typically spans two winter testing windows. The effective testing cycle for a single winter is usually only 3 to 4 months, with the standard window concentrated between November and March of the following year.
In other words, the full-vehicle development cycle can be compressed from 36 months to 18 months, and then to 12 months, but winter still occurs only once a year.
This is precisely where the "3 to 4 months left for suppliers" becomes truly critical.
When accelerated decision-making, early supplier involvement, parallel development, and AI simulation have all been utilized, yet the launch milestones are still at risk of being missed, the final phase to be further compressed is likely validation.
Therefore, AI boosting R&D efficiency and the rising pressure on validation are not contradictory.
AI can help identify issues early and reduce certain repetitive tasks, but it cannot make all physical validations—such as extreme cold, high temperatures, durability, and complex road conditions—disappear into thin air.
This issue is now receiving a response at the regulatory level.
Earlier this year, the Ministry of Industry and Information Technology (MIIT) repeatedly urged automakers to strengthen production consistency, reliability, durability, and the testing and validation of new technologies, explicitly stating that products lacking sufficient testing and validation must not be brought to market. In August, four ministries and commissions launched a one-year special campaign to improve production consistency and quality for road motor vehicle products, with reliability, durability, and new technology testing and validation listed as key priorities.
Revisions to relevant standards this year have sent the same signal: the reliability driving test mileage for NEVs (New Energy Vehicles) is proposed to be increased from the previous approximately 15,000 km to no less than 30,000 km.
These changes indicate that the shortening of development cycles is no longer just an efficiency issue; it is also becoming a quality management concern.
For suppliers, this shift is also redistributing value.
Suppliers in chassis, thermal management, mechanical, and safety-related fields, which require extensive physical validation and long-term durability testing, face greater time pressure; conversely, the value of companies that can provide AI R&D tools, simulation platforms, digital validation, and possess early collaborative development capabilities is being amplified.
Therefore, the real discussion surrounding the 12-month development cycle is not just about "whether it can be even faster."
Rather, it is about which time can be reclaimed through technology, and which validations must remain untouched under any circumstances.
As Supply Chain Boundaries Expand Beyond the Automotive Sector
If AI is changing "how vehicles are built," another transformation is occurring in "who builds the vehicle."
In his speech, Wang Xianbin mentioned that the next phase of innovation in the intelligent cockpit will likely originate increasingly from outside the traditional automotive industry.
Over the past few years, features such as large screens, voice assistants, seat adjustments, and configurations colloquially summarized as "refrigerators, color TVs, and large sofas" have been rapidly popularized.
As these features gradually transition from differentiating selling points to standard configurations, the next round of cockpit competition is expanding into health, comfort, space, and lifestyle.
Yet these capabilities are not necessarily held by traditional automotive Tier 1 suppliers.
Wang Xianbin revealed that over the past two years, the Gasgoo Research Institute has encountered numerous cross-industry supply chain directions in its front-end technology intelligence work.
For instance, can bedding and comfort materials from the hotel industry enter the automotive cockpit; can materials originally serving the female consumer market and sports health scenarios be redesigned for in-car use; and which mature technologies from industries such as massage and health monitoring can be redefined and integrated into vehicles.
This implies that the sources of innovation for next-generation cockpits may no longer be confined to the existing automotive supply chain.
This cross-industry integration is even more pronounced in the robotics sector.
On September 15, at the Gasgoo Auto Embodied Perception Fusion and Multimodal Large Model Innovation Seminar, Wang Xianbin presented a set of data: the overlap between automotive and robotics supply chains has already reached 60% to 70%.
In the same speech, he also noted that the execution system accounts for nearly half of a robot's BOM cost, and mainstream perception solutions heavily reuse the computing, sensing, and control systems already employed in vehicles and intelligent driving; as scale expands, there remains significant room for cost reduction across products with different positioning.
This indicates that the connection between automobiles and robots is no longer just about "technical similarity."
The supply chains themselves are overlapping.
Furthermore, at the AI-Defined Vehicle Forum on September 17, Wang Xianbin proposed an even more imaginative concept: could a robot dog become an AI Box within a vehicle?
According to his vision, when inside the vehicle, the robot dog could handle part of the edge computing workload; once the vehicle reaches its destination, it could detach from the car and continue executing tasks in scenarios such as camping, luggage handling, and child companionship.
In this way, vehicles and robots would no longer be entirely independent hardware entities but could share a unified set of AI, perception, and computing capabilities.
This concept is evidently still a long way from true large-scale commercialization.
The rapid shortening of the vehicle development cycle in China stems primarily from changes in the development mechanism.
However, the industrial logic behind it is beginning to become clear.
On the one hand, the sensors, controllers, motors, materials, thermal management systems, computing platforms, and mass manufacturing capabilities accumulated by automotive suppliers can be transferred to robotics; on the other hand, companies from industries such as hospitality, health, consumer electronics, and robotics are beginning to enter the automotive sector in reverse.
The boundaries of the automotive supply chain are shifting from internal restructuring to external expansion.
Wang Xianbin summarized the next phase of automotive development as an evolution from SDV—Software-Defined Vehicles—to AIDV—AI-Defined Vehicles.
However, based on the information revealed in this speech, what AIDV truly changes is not merely the addition of a large model to the in-car infotainment system.
On one end, it penetrates deeply into the R&D system, transforming how a vehicle is developed, validated, and delivered; on the other end, it continuously expands the boundaries of the supply chain, bringing in more technologies and companies that previously had no direct connection to the automotive industry.
The 12-month development cycle, the 3- to 4-month supplier validation window, and seemingly entirely disparate topics such as hotel materials, robotics, and AI Boxes ultimately point to the same transformation:
AI is redefining both the timeline and the boundaries of the automotive industry.