China's AI Model Usage Surpasses the US for the First Time: Kimi and DeepSeek Cases Highlight Open-Source Low-Cost Driving Effect

Factual Context

Public discussions reveal that China's AI model usage has surpassed that of US models. This phenomenon is linked to models such as Kimi. NVIDIA CEO Jensen Huang stated in an interview that Wall Street had previously misunderstood Chinese AI models, and after the DeepSeek case, a similar situation emerged with Kimi. He noted that open-source, lower-cost Chinese models will make AI usage more widespread, increasing adoption and thereby driving demand for graphics cards and data centers, benefiting NVIDIA.

This information comes from direct statements on social media, with no additional sources providing specific usage data or time details. Kimi is categorized as an open-source model and discussed alongside DeepSeek.

Mechanism Breakdown

Why has usage shifted? The core lies in open-source and cost advantages. Huang's remarks indicate that low-cost models lower the barrier to entry, making it easier for businesses and individuals to access AI services. As usage increases, demand for computing infrastructure grows accordingly. Graphics card procurement and data center construction become direct beneficiaries.

This operational model differs from that of closed-source models. Closed-source models typically rely on specific cloud services, while open-source models allow local or private deployment, reducing external dependencies. The materials do not specify deployment methods but emphasize that "open-source affordability" directly links usage expansion to hardware demand.

This sparks discussion on the US-China AI competition. The surpassing in usage may reflect differences in model accessibility rather than a single technological lead. Geopolitical factors are mentioned, but no quantitative evidence is provided in the materials, only serving as background discussion.

Industry Impact

For the competitive landscape, increased usage of Chinese models could alter the global distribution of AI services. The developer community engages actively, with discussions in both Chinese and English, indicating expanded model reach. Open-source characteristics allow more developers to participate in iteration rather than relying on a single vendor.

For developers, low-cost models provide more room for experimentation. The materials do not specify developer numbers, but rising usage suggests more application scenarios. For enterprise users, data security considerations may drive private deployment choices. Huang's perspective emphasizes that such models do not reduce hardware demand but instead stimulate procurement.

Across the overall supply chain, infrastructure providers receive positive signals. NVIDIA, as a major graphics card supplier, sees its demand expectations adjusted accordingly.

Strategic Assessment (Analysis)

Based on available materials, the most likely development is that open-source Chinese models will continue expanding usage scenarios, thereby maintaining or increasing hardware procurement demand. Huang's judgment anchors the link between "significant usage increase" and "stimulating demand." If this logic holds, it will influence supply chain planning.

Information still to be confirmed includes specific usage statistics and details on subsequent model iterations. The materials do not provide percentages or rankings, making it impossible to determine long-term leadership. Geopolitical discussions may persist, but the core drivers remain cost and open-source characteristics.

Developers and enterprise users may increasingly shift toward hybrid deployment models, using open-source models to reduce costs while leveraging local hardware for security. Infrastructure companies like NVIDIA may adjust product strategies accordingly, focusing on supporting large-scale inference scenarios.

Overall, the event reflects the diversity of AI adoption pathways rather than a single dominance. Future observation should focus on whether usage changes translate into sustained hardware orders.