Is the launch speed of new cars actually faster than that of smartphones now?
In the first half of 2026, the domestic passenger vehicle market ushered in an explosion of new car launches. According to comprehensive statistics from the Dongchedi New Car Launch Database, a total of approximately 630 new car models were launched in China from January to June this year, averaging over 3 models per day. Even based on the vehicle series caliber of the Yiche Vehicle Model Database (excluding different trim levels under the same series), there were 576 new vehicle series and 1,731 vehicle models launched in the first half of the year, averaging over 3.3 models daily.
For comparison, data from the China Academy of Information and Communications Technology (CAICT) shows that a total of 185 new mobile phone models were launched in China from January to June 2026, averaging about 1 model per day. It is evident that the pace of new product launches in the automotive industry has already surpassed that of the consumer electronics industry, which is known for its fast pace.
With new car launches outpacing smartphones, how exactly is the R&D speed sustained? In particular, now that intelligent driving systems have become standard equipment, has the R&D process actually become simpler, or has it become more complicated? Today, let's discuss this topic.
01. From Scratch to Finished Product: What Processes Does a Car Go Through?
The complete R&D process for a new car can be roughly divided into five stages: strategic planning, concept design, engineering development, testing and validation, and production preparation. A review gate is set at the end of each stage, and only projects that pass the review can proceed to the next stage.
The strategic planning stage determines what kind of car to build. Automakers need to complete market analysis, benchmark against competitors, and develop user personas to clarify the vehicle positioning, powertrain type, performance metrics, and cost boundaries. Once the project is officially approved, core suppliers are locked in to participate in early technical alignment.
The concept design stage addresses what the car will look like. Designers will filter directions from a large number of sketches, refine the styling using 3D digital modeling, and then create a 1:1 clay model to confirm the surface lighting and shadow effects as well as the tactile feel of the details. The clay model generally undergoes multiple rounds of review, because once the engineering development stage begins, the room for styling adjustments becomes very limited. In the overall vehicle development cycle, it usually takes about one to one and a half years from the initiation of styling design to the freezing of the design scheme.
The engineering development stage solves the problem of how to actually build the car. The design of the body structure must meet collision safety requirements, while the chassis and suspension systems need to find a balance between handling and comfort; bench calibration of the powertrain is also advancing in parallel to ensure that energy consumption and power output meet target values; the development of software and the Electrical/Electronic (E/E) architecture is also carried out in this stage, with the schemes for the driver assistance system, intelligent cockpit, and whole-vehicle OTA (Over-The-Air) capabilities being finalized.
After the design of each subsystem is completed, prototypes will also evolve progressively according to their maturity, starting from A-samples, going through B-samples and C-samples, until D-samples complete the final validation and release mass-production parts. Only after all designs are frozen and validated can the project proceed to the next stage.
Testing and validation is the most time-consuming phase. For a completely new vehicle model, from design freeze to mass production delivery, the conventional safety validation cycle takes at least 18 to 24 months. The whole vehicle must undergo virtual simulation testing and global road field tests. The "Three Highs" tests—high temperature, extreme cold, and high altitude—are all indispensable. Tens of millions of kilometers of road testing and hundreds of collision simulations are completed in this stage. Taking GAC (GAC Group) as an example, its new cars must undergo an 18-month validation cycle covering two winters and one summer from project initiation to launch, with a conventional development cycle of 21 months.
If issues are found during the validation process, the engineering team needs to return to the design stage for modifications and then conduct another round of validation. At the component level, reliability validation testing after a single design change takes at least 1 to 2 months, and system-level validation at the OEM (Original Equipment Manufacturer) level also requires 1 month. This cycle of validation, issue discovery, modification, and re-validation is the fundamental reason why the testing and validation stage takes the longest.
The production preparation stage involves completing the development of tooling and molds, production line debugging, and trial production. Only after confirming that everything is ready can mass production begin.
This process is the standard practice accumulated by the industry over decades. However, in the era of intelligent electric vehicles, the R&D cycle has instead been significantly compressed. Traditional internal combustion engine (ICE) vehicles required 36 to 48 months, whereas mainstream NEV (New Energy Vehicle) manufacturers have now compressed it to 18 to 24 months. With the shortened cycle, whether the completeness of the validation process can be guaranteed has become a question worth asking.
02. With the Arrival of Intelligent Driving Systems, Validation Is Actually More Complex
Testing and validation for traditional vehicles mainly revolve around mechanical and powertrain systems, including chassis tuning, powertrain calibration, and durability testing. The addition of intelligent driving systems has introduced an extremely complex validation layer to this framework. The number of sensors, controllers, and actuators in the whole vehicle has surged, and the hardware-software interfaces have grown geometrically. The traditional R&D model, which relies on manual experience and serial validation, is approaching its efficiency ceiling.
For the validation of an autonomous driving algorithm, the industry's common practice follows a rough ratio: approximately90% of the workload is completed in a simulation environment, about 9% in closed-course testing, and about 1% on public roads.
Simulation testing requires running massive amounts of extreme scenarios, such as e-bikes suddenly darting out on urban roads, blurry lane markings in heavy rain, and rural intersections without traffic lights. These are all completed in simulation scenarios. The reason for this allocation is that these extreme scenarios are difficult to encounter on real roads but can be tested repeatedly in a simulation environment.
Closed-course testing is mainly used to verify the physical consistency of simulation results, while public road testing is used for final compliance validation and regulatory filing.
In January 2026, the recommended national standard "Intelligent Connected Vehicles—Simulation Test Methods and Requirements for Autonomous Driving Functions" (GB/T 47025-2026) was officially implemented. The standard covers all aspects of simulation testing for autonomous driving functions, from test requirements, test methods, overall pass requirements, to simulation credibility evaluation.
The standard requires that the error between the sensor model and the actual vehicle does not exceed 5%, and the consistency between the dynamics model and the actual vehicle is no less than 95%. Simulation testing is no longer just an auxiliary means for R&D, but a formal validation step that must achieve a certain level of credibility.
Following simulation are closed-course testing and public road testing. The requirement for Volkswagen Group is that every mass-production model equipped with ADAS (Advanced Driver Assistance Systems) must complete over 300,000 kilometers of real-road validation before launch. This includes 70% daytime driving and 30% nighttime driving, as well as 50% urban roads, 30% highways, and 20% suburban roads. The validation process includes two rounds of extreme environment testing in summer and winter.
Of course, some brands in the industry only conduct one round of extreme environment validation.
In June 2026, the Ministry of Industry and Information Technology (MIIT) released the first mandatory national standard for L2 combined driving assistance, "Intelligent Connected Vehicles—Safety Requirements for Combined Driving Assistance Systems" (GB 47955-2026). The standard specifies the safety requirements for pilot combined driving assistance systems, basic single-lane and multi-lane combined driving assistance systems.
According to the standard requirements, the industry needs to establish a three-level validation mechanism combining simulation, closed courses, and real vehicle roads. Algorithm iterations and OTA upgrades must be accompanied by complete simulation validation materials.
In addition, mandatory standards for L3 and L4 levels are also being advanced. In June 2026, the MIIT published the draft for approval of the mandatory national standard "Intelligent Connected Vehicles—Safety Requirements for Autonomous Driving Systems." The standard requires that the system safety level must not be lower than that of a qualified and attentive human driver, and establishes a safety dossier mechanism integrating declarations, arguments, and evidence.
The implementation of these standards means that the validation of intelligent driving systems is no longer an optional action for automakers, but a mandatory one. A complete intelligent driving validation process, from simulation to closed courses and then to open roads, takes no less time in total than traditional durability testing.
03. Where Is the Compressed Time Squeezed From?
Intelligent driving systems have added validation steps, so why has the R&D cycle instead been compressed from 55 months to 18 months? In fact, to accelerate the launch of new cars, many automakers squeeze time in stages.
The strategic planning and concept design stages can be accelerated through AI (Artificial Intelligence) assistance and concurrent engineering. Ivan Espinosa, President of Nissan Motor, announced in June 2026 that the company has compressed the new car R&D cycle from 55 months to 26 months. AI and digital tools have been widely applied in the design, testing, and manufacturing stages, with some physical prototype testing completed via computer simulation. Nissan stated that virtual simulation testing has replaced over 60% of real vehicle testing.
The engineering development stage can be compressed through platformization and modularization; however, the room for compression in the testing and validation stage is limited. A car must drive the required mileage in an environment of minus 40 degrees Celsius, and complete sufficient cycle tests in high-temperature and high-altitude environments. There are no shortcuts for these tasks. Although virtual simulation can replace a part of it, it cannot replace it entirely.
In January 2026, the MIIT revised and released the new version of access review requirements, upgrading reliability testing from internal corporate control to a national mandatory requirement. However, the compression of the R&D cycle is putting these validation steps at risk of being squeezed. An R&D insider revealed that vehicle control software, which originally required 4 months of validation, can now be put into production in just two weeks; projects that originally required 200 tests might end up with only 30 tests.
When the R&D cycle of a car is compressed to the extreme, steps that should be carried out without compromise, such as extreme environment road testing and durability validation, are easily omitted. Hidden dangers such as software defects, battery thermal runaway, and assisted driving failures follow suit.
Li Shufu, Chairman of Geely Holding Group, publicly stated in June 2026 that automotive R&D can be accelerated, but testing steps must never be reduced, and the safety bottom line cannot be compressed. These remarks point directly to the widespread problem currently facing the industry: when everyone is competing in the speed of launching new products, the compressed validation time may ultimately be paid for by user safety. The MIIT elevating reliability testing to a mandatory requirement is, to some extent, a correction to the excessively fast pace of the industry.
04. The Speed of New Product Launches and Market Returns Have Severely Diverged
Comprehensive statistics from the Dongchedi New Car Launch Database show that a total of approximately 630 new car models were launched in China in the first half of 2026. This figure covers annual minor facelifts, configuration derivatives, and limited-edition exclusive models. If various newly added models are included, the total number of new car SKUs in the entire market reaches 3,780, which translates to an average of about 6 versions per car model.
The intensity of new car launches has hit a record high, but the market feedback is not optimistic. Statistics from the China Passenger Car Association (CPCA) show that domestic passenger vehicle retail sales totaled 8.701 million units in the first half of the year, a year-on-year decline of 20.2%. Looking at specific models, fewer than 30 models maintained stable monthly sales of over 10,000 units in the first half of the year. About 35% of new cars had monthly sales below 2,000 units, and some models sold as few as 36 units.
In addition, industry profits are also continuing to decline. Chen Shihua, Deputy Secretary-General of the China Association of Automobile Manufacturers (CAAM), publicly stated in July 2026 that the profit margin of the automobile manufacturing industry from January to May this year was only 3.4%, far below the national average of 5.56% for industrial enterprises above designated size. Data from both the CPCA and the National Bureau of Statistics show that the profit margin in the whole vehicle manufacturing segment has dropped to 1.5%, hitting a historical low, with the total profit of whole vehicle manufacturing decreasing by 43% year-on-year. In the first quarter of this year, the industry's gross profit per vehicle fell to 14,000 RMB, a drop of nearly 40% compared to the same period in 2025.
The launch density of over 3 new cars per day has been exchanged for a 20% decline in total market volume and a profit margin dropping to 3.4%. A large number of new cars have suffered dismal sales after launch, making it impossible to recover early investments, and the loss pressure across the industry continues to intensify.
The speed of new car launches has already surpassed that of smartphones, which clearly exceeds the carrying capacity of the traditional framework for automotive R&D. Especially now that intelligent driving systems have become standard equipment, the validation process should have become more complex. Even if digital tools and platform-based development can compress part of the cycle, this compression must have its limits.
For automotive R&D, no matter how useful new technologies are, not a single step of the required testing road can be skipped.
#AutonomousDriving #AutomotiveRnD #ShortenedAutomotiveRnDCycle