Farewell to the "concept war", mass production capacity becomes the key to victory for assisted driving enterprises.

Entering 2025, the automotive industry has witnessed a storm of "equal access to assisted driving". In the past, only vehicles priced over 300,000 yuan were equipped with advanced assisted driving features. However, due to the intense competition among automakers, these features have now been introduced to models priced around 100,000 yuan, truly marking the beginning of the era of assisted driving popularization.

The achievement of equal access to assisted driving is primarily due to the continuous decline in the prices of core components. For example, the price of a single lidar has dropped from tens of thousands of yuan to around a thousand yuan in the past decade. Moreover, technological advancements have played a crucial role. Currently, there are corresponding solutions for different application scenarios of high, medium, and low levels of assisted driving. At the 2025 Shanghai Auto Show, Nullmax showcased its self-developed full-stack MaxDrive assisted driving solution, which covers all scenarios from basic ADAS to advanced urban NOA, meeting the diverse needs of different vehicle models and markets. This demonstrates Nullmax's "universal value" in providing sustainable long-term development for the entire industry.

However, problems have also emerged. For automakers, the implementation of technology may not be the most challenging task. The real challenge lies in finding a sustainable business model to monetize the technology. In Nullmax's view, although the autonomous driving industry is developing rapidly, there are still many diverse needs in the high, medium, and low - end markets that need to be met, and different commercialization challenges are faced.

I. The Underlying Logic of Universal Assisted Driving

At the 2025 Shanghai Auto Show, Nullmax, an assisted driving application company, presented its latest technological achievements, including the MaxDrive intelligent driving solution (with four versions) and the new - generation end - to - end assisted driving technology.

First, there is MaxDrive Basic, an intelligent front - view integrated machine solution. This is an algorithm platform solution developed for L2 - level assisted driving. Simply put, this solution is designed for the entry - level market, where the demand is more focused and cost - sensitive. Entry - level users do not have as segmented and personalized needs for assisted driving as high - end users. For example, most users only require basic assisted driving functions such as ICC intelligent cruise control and mainstream ADAS. The MaxDrive Basic solution can meet the needs of these users.

Meanwhile, the intelligent front - view product included in this solution can achieve basic assisted driving based on the Texas Instruments AM62Ax chip with 2T computing power. This effectively reduces costs and can meet the diverse needs of both overseas and domestic markets, helping partners achieve rapid mass production.

In the mid - to - high - end field, Nullmax has introduced MaxDrive Standard, a parking and driving integrated domain control solution, which is an L2+ assisted driving solution. It can achieve functions such as high - speed NOA, memory driving MD, memory parking HPA, and automatic parking APA, covering almost all the advanced functions that assisted driving should have.

On this basis, a new trend has emerged in the market: cockpit - driving integration. This design concept not only optimizes the utilization of computing power but also significantly enhances the experience of assisted driving functions and effectively reduces costs. It is reported that Nullmax's MaxDrive Standard_Hyperlin cockpit - driving integration solution has been successfully adapted to the Qualcomm SA8775 (48 TOPS) and Black Sesame Technologies C1296 platforms (52.5 TOPS) and meets the 5 - star regulatory requirements of GSR 2.0, CNCAP, and ENCAP. Through extreme optimization of computing power utilization and reasonable and efficient resource allocation, this solution squeezes out every bit of computing power from the chips. Among them, the solution based on the Qualcomm SA8775 chip combined with NXP S32G can support multi - domain integration, connecting intelligent driving, cockpit, gateway, BCM, VCU and other modules. In terms of driving assistance functions, it can achieve advanced ADAS functions including urban commuting (MD) and memory parking (HPA). In the future, through Nullmax's platform - based technology, this solution will transplant AI algorithms to Renesas X5 series chips, truly empowering the major global automotive markets in China, Europe, America, Japan, and South Korea.

Entering 2024, a new end - to - end storm has swept through the assisted driving circle. It has become a trend that if you don't use large models, you can hardly claim to be involved in assisted driving. Nullmax has launched another version, MaxDrive Plus: an end - to - end multi - modal large - model assisted driving solution, which is a high - end assisted driving solution for full - scenario NOA (Highway + Urban). This solution has a higher technological content. By adopting an advanced visual language model (VLA) and an end - to - end architecture, it can achieve intelligent driving in complex urban road environments without relying on high - precision maps and lidars.

He Xiaopeng, the chairman of XPeng Motors, once sighed that developing assisted driving is extremely costly, and it is difficult to sustain R & D without continuous and stable profits. With the MaxDrive Plus solution, automakers can save the costs of high - precision maps and lidars and develop advanced assisted driving at a low cost.

The above four versions cover the needs for assisted driving in different scenarios of high, medium, and low levels. The high compatibility is achieved because Nullmax has independently developed a platform - based technological architecture. With this architecture as the foundation, full - scenario coverage of assisted driving can be realized, from basic ADAS to urban NOA. It supports different applications such as end - to - end multi - modality, parking and driving integration, cockpit - driving integration, and front - view integration, covering all driving scenarios including urban areas, highways, and parking, and providing a smooth upgrade path from L2+ to L4.

Zhang Fan, the senior director of Nullmax's mass - production engineering, explained to 36Kr the thinking behind Nullmax's insistence on developing a platform - based architecture. "Building a universal assisted driving platform is of great significance for Nullmax, a full - stack assisted driving algorithm company centered on visual perception and based on deep learning."

This is because Nullmax's customers come from different regions around the world, and the demands of the end - markets vary. Zhang Fan gave an example: "Some customers need to develop advanced urban functions to handle various complex scenarios, some pursue point - to - point functions on structured roads (such as highway scenarios), and some only need front - view integrated machine functions for regulatory scoring." Different requirements mean different sensor configurations. Designing algorithms for each case would result in a huge workload and high costs, and the results may not be the best.

The platform - based architecture can effectively solve the above problems. Similar to the platform - based vehicle manufacturing model commonly adopted by automakers, it can adapt to different solutions to the greatest extent at a lower cost.

Moreover, Nullmax's AI algorithm platform - based technological architecture enables its assisted driving solutions to flexibly support different chips and can be efficiently adapted to the current mainstream assisted driving chips at home and abroad. This advantage lies in not being restricted by others and having more choices when selecting suppliers.

However, the development of this platform - based technological architecture has not been smooth sailing. In the AI era, the biggest challenge is how to quickly obtain high - quality labeled data to solidify a basic model with high compatibility and robustness.

This is also the case for Nullmax. Zhang Fan sighed, "There is no shortcut in this regard. We have to invest manpower and resources to build Nullmax's data closed - loop system step by step, create a cloud - based data platform integrating data collection, labeling, and training, and create good pre - conditions for training a good basic model."

At the same time, Nullmax has also made many efforts to make the cloud - based data platform operate more efficiently. For example, to improve the labeling efficiency and reduce the labeling cost, Nullmax uses large models for automatic labeling; to save the cost of data collection, Nullmax uses AIGC to generate many corner cases that are difficult to encounter in real - world scenarios with high fidelity; in addition, to improve the search efficiency of effective data, Nullmax's platform also supports multi - modal search methods such as text - to - image and image - to - image search, which greatly facilitates the development of engineers.

II. Only Commercially Viable Assisted Driving is Good

Compared with the current prosperity, industry insiders still have lingering fears about the autonomous driving cold wave in 2022.

In August of that year, Aurora, known as the "world's first autonomous driving stock" and one of the three major players in the autonomous driving field along with Waymo and Cruise, was reported to be in business difficulties. It had to cut costs by laying off employees, reducing executive salaries, selling assets, and reducing welfare.

Even earlier, Tesla laid off nearly 200 positions in its autonomous driving department in California; Cruise, an autonomous driving subsidiary of General Motors, announced the dismissal of nearly 8% of its employees; and Argo AI, backed by Ford and Volkswagen, could not hold on. This company, once valued at $7 billion and regarded as a star enterprise in the L4 - level autonomous driving field, suddenly announced its closure.

The root cause of this cold wave is quite simple. Technological R & D requires continuous and huge capital investment, but the players in the industry have not yet found a sustainable business model. This has also led people to start thinking about how to achieve the commercialization of autonomous driving as soon as possible.

Nullmax's approach is that autonomous driving should develop gradually. Zhang Fan explained that they do not believe that autonomous driving will reach the L5 level overnight. It will definitely be a long - term process. As an AI technology company, Nullmax will work with the industry to achieve full - scale autonomous driving as data is gradually improved and model capabilities are gradually enhanced.

Based on this approach, Nullmax has taken the lead in achieving large - scale mass production applications in the pre - installation field. It has established in - depth mass - production - oriented cooperation with many well - known domestic and foreign automobile manufacturers and first - tier suppliers and has won mass - production projects from several domestic automobile manufacturers, including SAIC, Chery, Jiangling, Desay SV, and Jingwei Hengrun.

In addition, Nullmax's systematic solutions, visual perception technology, and multi - core heterogeneous development capabilities are highly recognized in the industry. It has achieved several firsts and onlys in the industry and has won the favor of global automobile manufacturers, Tier 1 suppliers, and chip manufacturers for cooperation, jointly promoting the large - scale implementation of autonomous driving.

This undoubtedly provides strong support for Nullmax to develop more complex autonomous driving technologies and gains valuable time.

The gradual approach continues to influence Nullmax.

At different development stages, identifying the different needs of users and providing matching products is a more suitable strategy for the healthy development of the enterprise.

Nullmax's parking and driving integrated solution and some of its technical modules have won multiple mass - production contracts. In terms of the time of commercialization and the number of orders, it is at the forefront of the industry. In recent years, Nullmax has completed mass - production deliveries for multiple customers and received good feedback. Currently, Nullmax's product solutions have been adapted to multiple chips and deployed in real vehicles, with the number of supported chips and brands ranking among the top in the industry. From 2024 to 2025, it has won projects from several leading automakers/Tier 1 suppliers and started large - scale mass - production deliveries.

Zhang Fan revealed that in addition to the gradual approach, Nullmax's good commercial progress also benefits from the compatibility and robustness of its core algorithms. Moreover, engineering is also an essential part of commercial implementation. "In the past and ongoing mass - production projects, our engineering team, customers, and partners work together in a closed - loop development mode. Although this process is painful, it is necessary. Through this process, we can better understand the customers' needs and solve problems more efficiently."

III. Nullmax's New Card: Assisted Driving Goes Global

It is undeniable that although the origin of autonomous driving is overseas, with the rapid development of new energy vehicles in the Chinese market, the competition in assisted driving has intensified in China. This has become a major advantage for many new energy automakers going global.

Going global is also Nullmax's next plan.

As early as the 2024 CES, Nullmax jointly presented a diverse range of solutions and mass - production technologies with TI.

Based on TI's Jacinto TDA4x processor and Sitara AM62Ax processor, Nullmax showcased a series of solutions, including the world's first 8T computing power parking and driving integrated mass - production application with NOP and HPP functions and the world's first intelligent front - view product based on 2T computing power.

The solutions presented at that time could meet the development needs of both new domestic models in China and new overseas models, complying with the requirements of different regions in terms of system functional safety, information security, and chip safety.

Currently, Nullmax has reached cooperation intentions with several global partners, and its products will be launched overseas in the near future. Nullmax participated in the CES in 2024 and 2025. From advanced assisted driving to basic ADAS, it jointly presented a number of exclusive products with industry partners. These products have rich functions, excellent performance, and cost advantages, meeting the different needs of global automakers and consumers from "ultimate experience to ultimate cost".

However, the road to going global is not always smooth. When assisted driving goes overseas, it needs to face relevant regulations in different regions, and each region has complex and strict access regulations. For example, the new GSR regulations in the EU (Regulation (EU) 2019/2144 issued in 2019) have required all vehicles registered in the EU to meet new standards since July 2024, adding a number of safety technology requirements, such as the mandatory installation of intelligent speed limiters (ISA) and emergency braking systems (AEB), and other four major categories of advanced safety functions. Although the E - NCAP standard is not mandatory, it is authoritative and clearly requires various active safety functions of passenger cars.

Nullmax has also felt this acutely. Zhang Fan introduced that as an algorithm company that attaches great importance to safety, Nullmax has a development system that strictly follows ISO 26262 and 21448, and has established a complete development process from the system to the software. This ensures that every line of code and every function is safe enough to help overseas mass - production products meet the highest standards of E - NCAP and GSR.