Gasgoo Munich- Jensen Huang's 33-year tenure as CEO came to an end on November 12.
As Huang departs, rivals and industry players are already recalibrating their positions, while the battle over AI chips in Washington intensifies.
OpenAI has told U.S. lawmakers it backs a bipartisan push to tighten chip export restrictions, Bloomberg reported. Anthropic has publicly voiced support for similar measures. Industry insiders view this as a targeted strike.
NVIDIA employees have petitioned en masse, calling for Huang's return, leaving the outcome uncertain. But what does this mean for the Chinese market?

Image Source: Doubao
Short Term: Can't Be Replaced
"NVIDIA chips hold a unique competitive position and can't be replaced anytime soon," several industry insiders told Gasgoo.
Over the past few years, Chinese automakers have poured money into high-level autonomous driving, making NVIDIA a key supplier for premium intelligent driving computing platforms.
Data from Gasgoo Automotive Research Institute shows NVIDIA held a 45% share of China's smart driving domain-control chip market from January to September 2026, ranking first. Combined installations of its Orin-X, Orin-N, and Thor-U products reached 2.31 million units. Despite tightening U.S. export controls, NVIDIA remains the leader in China's intelligent driving chip market.
Purchasing volumes highlight that dependency. Nio founder and chairman William Li revealed the company bought about 800,000 NVIDIA chips in 2024, worth roughly $300 million. For an automaker, that's a significant outlay.
NVIDIA's edge isn't just raw computing power. Intelligent driving requires processing massive amounts of sensor data to recognize environments and make driving decisions.
Orin and Thor deliver in-vehicle computing power, while software tools like DriveOS, CUDA, and DriveWorks form the underlying development environment for algorithm deployment. Meanwhile, NVIDIA uses Omniverse to reconstruct real-world road scenes and build high-fidelity simulations, and leverages Cosmos to generate complex driving scenarios — all supporting the training and validation of autonomous driving models.
Automakers use these tools to test complex road scenarios and verify algorithm performance before deploying systems to vehicles.
For automakers, choosing NVIDIA means adopting a development ecosystem validated over years. Some even emphasize NVIDIA chips at new-car launches, borrowing the brand's technical credibility.

Image Source: Generated by Doubao, compiled by Gasgoo
That's why industry insiders believe NVIDIA's procurement business in China won't be affected by management changes in the short term. Which Chinese Automakers Use NVIDIA?
NVIDIA's partnerships with Chinese automakers now span multiple brands and product platforms.
Nio was among the earliest Chinese automakers to adopt NVIDIA's computing platform. In 2021, Nio announced a partnership to use DRIVE Orin chips in its next-generation intelligent EVs. Its in-house Adam supercomputing platform, unveiled that same year, centers on four DRIVE Orin SoCs, delivering over 1,000 TOPS of computing power. Since then, mass-produced models like the ET7, ET5, and ES7 have all featured this platform.
BYD also announced in 2023 that it would expand Orin's use across its new energy lineup, covering models in the Dynasty and Ocean series.
In 2024, NVIDIA's partnerships with Chinese automakers expanded further. Li Auto revealed it would use DRIVE Thor as the central computing platform for next-generation models, while Great Wall Motor, Zeekr, and Xiaomi adopted Orin to develop intelligent driving systems. That same year, BYD, XPENG, and GAC HYPTEC announced plans for Thor collaboration. NVIDIA's partner base has gradually shifted from primarily high-end smart EVs to a broader range of domestic brands and product lines.

Image Source: Generated by Doubao, compiled by Gasgoo
This expansion reflects shifting dynamics in China's auto market. As city-assisted driving becomes a critical competitive battleground and the scale of new energy vehicles grows, these features are trickling down from premium models to more mainstream price segments. NVIDIA platforms are following suit.
By 2026, cooperation had deepened further. Early that year, BYD and Geely announced they would use NVIDIA's DRIVE Hyperion platform to develop Level 4 autonomous vehicles. Hyperion integrates in-vehicle computing, sensors, networking, and safety systems, extending collaboration to the entire autonomous vehicle architecture.
Shortly after, Chery announced a global strategic partnership with NVIDIA to collaborate on assisted driving, cockpit AI, and robotics. Chery's L3 and L4 models for the Chinese market will adopt the DRIVE Hyperion platform and incorporate technologies like Alpamayo and Cosmos.

NVIDIA's partners now extend beyond automakers to autonomous driving solution providers and supply chain companies. WeRide and Pony.ai have begun developing solutions based on DRIVE Hyperion, while suppliers like Desay SV are participating in NVIDIA's automotive ecosystem.
Publicly, NVIDIA has established two interconnected paths in China. One involves embedding Orin and Thor into specific models for current and next-generation products. The other revolves around Hyperion for L3 and L4 platform development, allowing NVIDIA to participate early in automakers' future technical planning.
As the chart shows, as intelligent driving competition spreads across price points and automation levels, collaboration with NVIDIA is shifting from individual vehicle configurations to platform development. For NVIDIA, Chinese automakers remain crucial clients, with the value of partnership increasingly tied to R&D for next-generation intelligent vehicles.
NVIDIA in China Is About More Than Just Cars
If cars are NVIDIA's gateway into the physical world, robotics and industrial manufacturing connect it to a wider array of Chinese enterprises.
Robotics is a key growth area for NVIDIA. The humanoid robot industry is now entering production ramp-up and scenario verification. Domestic shipments exceeded 40,000 units in the first half of 2026, with full-year production expected to top 100,000. Yet, looking at sales structure, research, education, exhibitions, and government data centers remain the primary demand sources, accounting for roughly 70% of shipments.
There is still room for growth in commercial applications that handle actual industrial tasks. That presents an opportunity for NVIDIA, as robot makers need extensive simulation, data generation, and model training tools during R&D.
NVIDIA's Isaac platform provides the tools for this process: Isaac Sim simulates real environments, tests robot behavior, and generates synthetic data, while Isaac Lab trains robot motion policies.
Building on that, NVIDIA has extended its technology into robotics foundation models, launching GR00T and updating it to GR00T 2 in November 2026. The new model can simultaneously predict future scenes and robot actions, allowing robots to learn skills from human videos and minimal demonstrations, reducing reliance on massive amounts of real-world training data.

Image Source: Generated by Doubao, compiled by Gasgoo
This technology stack is already entering the R&D processes of Chinese robotics companies. Fourier Intelligence, Galaxy General, and XPENG Robotics have applied the Isaac platform to simulation, training, and development. Additionally, NVIDIA provides domestic robot firms with technical support ranging from computing hardware to development software via its Jetson edge computing platform and Isaac Sim tools.
NVIDIA's business thus extends from simulation and training in the R&D phase to the computing required for actual deployment.
Industrial manufacturing is another key expansion path. While robots focus on individual units and tasks, manufacturers must consider entire factory workflows, equipment coordination, and logistics. Retrofitting a factory involves countless devices and production steps, where any layout change can disrupt subsequent construction and production. Companies therefore need to verify designs as thoroughly as possible before breaking ground.
NVIDIA's Omniverse platform is designed for exactly this. Companies can digitally reconstruct factories, production lines, and equipment in a virtual environment to simulate workflows and validate equipment layouts, robot configurations, and logistics paths in advance.
BMW has used Omniverse to build a virtual factory. Its digital twin covers over 1 million square meters — roughly equivalent to 140 football fields. BMW can plan production lines, optimize robot configurations and logistics systems, and adjust designs in the virtual space, minimizing disruption to actual production.

Image Source: NVIDIA
NVIDIA's Cosmos further integrates physical world understanding and generation into AI model training. Supporting five modalities — text, image, video, environmental audio, and motion — the model can perform physical world reasoning, scene generation, and motion prediction.
For robot developers, Cosmos can generate data under various weather, lighting, and environmental conditions based on real-world scenes. This supplements training samples that are difficult to collect at scale, helping robots handle a wider range of complex scenarios.
From robot training to factory digital twins to physical world data generation, NVIDIA is linking computing hardware, simulation tools, and foundation models. While these technologies serve different industries, they share a common need: companies require continuous access to computing resources, software tools, and models when developing and deploying physical AI systems. NVIDIA aims to integrate itself deeper into the R&D and production workflows of more Chinese enterprises.
Is the Opportunity Here for Domestic Chips?
In the long run, Jensen Huang's departure will inevitably affect NVIDIA. As one industry insider put it, "When the spiritual totem falls, there's definitely an impact."
Huang hasn't just steered NVIDIA's product strategy in recent years; he has also personally engaged with key clients and built the industry ecosystem. His personal influence has become intertwined with the NVIDIA brand. His successor must maintain continuity in technology while expanding the business into data centers, automotive, robotics, and industrial computing.
However, Huang's departure doesn't imply NVIDIA's business system will change overnight. Chip development, software platforms, and client partnerships are supported by vast teams, and existing product roadmaps and mass-production projects won't shift immediately due to management changes. One robotics company told Gasgoo that its collaboration with NVIDIA remains unaffected.
For Chinese automakers, the most immediate short-term impacts come from product supply, export controls, and technological iteration. Models already developed around NVIDIA platforms involve chip adaptation, algorithm verification, and vehicle testing — making it difficult for automakers to switch technical solutions solely due to a management change.
Compared to management turnover, U.S. export controls have a far more direct impact on NVIDIA's China business. In fiscal 2026, revenue from mainland China and Hong Kong stood at $19.677 billion, down from $25.048 billion the previous year. Export restrictions not only hurt NVIDIA's Chinese operations but also force automakers to weigh supply stability more carefully when selecting chip suppliers.
This creates new room for domestic chips. In recent years, more Chinese automakers have integrated chip development into their own technology stacks. Nio's in-house Shenji NX9031 chip has entered mass production, while XPENG's Turing chip is now targeting automotive and robotics scenarios. Meanwhile, professional chip firms like Horizon Robotics and Black Sesame are pushing product iterations to secure spots in more automakers' mass-production projects.

Image Source: Generated by Doubao, compiled by Gasgoo
Chinese automakers are willing to invest in self-research for business reasons. As intelligent driving becomes a critical competitive factor, chips directly influence system performance, cost, and product iteration. Mastering chip design allows automakers to better coordinate hardware, algorithms, and vehicle development while reducing reliance on a single supplier. The sheer number of Chinese automakers and their rapid product cycles also provide more opportunities for verification and mass production.
But completing chip development is just the first step in entering the automotive supply chain. Automotive-grade chips require rigorous reliability and safety validation, followed by long-term testing once inside a vehicle. High-end intelligent driving chips must also work in tandem with operating systems, algorithm frameworks, and domain controllers, requiring chipmakers to provide ongoing development tools and software support. Compute power metrics announced at launch don't always translate directly to real-world vehicle performance.
This is the main hurdle domestic chips face when competing with NVIDIA. After years of accumulation, NVIDIA has built a relatively mature ecosystem of chips, software, and development tools. Automakers have also invested heavily in algorithm development and system adaptation around NVIDIA's platform.
While domestic companies have made progress with certain products, they still need to accumulate adaptation experience across different scenarios through mass production in more models, gradually building a stable software ecosystem and developer community.
Therefore, domestic chips gaining more mass-production opportunities doesn't mean NVIDIA will quickly exit the Chinese market. Automakers vary in R&D capability, product positioning, and cost requirements. Even a single automaker might adopt different chip solutions depending on the model's positioning, intelligent driving features, and development needs.
External procurement, in-house chips, and domestic supplier solutions are set to coexist for the foreseeable future.
For Chinese automakers, the real question is whether NVIDIA's position in the U.S.-China chip rivalry will shift. As a U.S. chipmaker, NVIDIA's business in China is perpetually subject to export control policies. It remains unclear whether a change in management will alter its China strategy.
That's why Chinese automakers are continuing their partnerships with NVIDIA while accelerating their own chip development and domestic supplier relationships. As domestic chips enter mass production, automakers are placing greater weight on supply stability, development costs, and technological autonomy. The real change domestic chips bring is giving automakers more leverage in chip selection — and subjecting NVIDIA to more sustained competition in China.








