For the third consecutive week, China's artificial intelligence sector has outpaced the United States in large model usage, a trend supercharged by a 210% weekly surge in the use of Tencent's Hy3 preview model.
For the third consecutive week, China's artificial intelligence sector has outpaced the United States in large model usage, a trend supercharged by a 210% weekly surge in the use of Tencent's Hy3 preview model.

For the third consecutive week, China's artificial intelligence sector has outpaced the United States in large model usage, a trend supercharged by a 210% weekly surge in the use of Tencent's Hy3 preview model.
Data for the week of May 11-17 shows Chinese AI models recorded 7.693 trillion token uses, nearly doubling the 4.24 trillion from U.S. counterparts, according to estimates from OpenRouter. The growth highlights a rapid developer adoption cycle in China that is challenging the long-held dominance of American firms in foundational AI, with Tencent's Hy3 preview model leading the global rankings.
The surge reflects a broader ambition within Chinese-affiliated technology companies to take a leading role in the next wave of AI. While not directly involved in large language models, Faraday Future's founder YT Jia recently articulated a vision of building an "EAI ecosystem" to become a "global EAI industry bridge," a strategy that mirrors the platform-based, multi-partner approach gaining traction in China's software sector.
Global AI large model token usage reached 26.9 trillion for the week, a 4.7% increase, marking four straight weeks of growth. Chinese models accounted for over 28% of that total. Tencent's Hy3 preview model alone registered 2.66 trillion tokens, a 210% week-over-week spike that placed it first in global usage. Another Chinese model, DeepSeek-V4-Flash, also ranked in the top three, showing the country's growing strength is not limited to a single provider. In contrast, U.S. model usage grew a more modest 12.77% week-over-week.
This sustained lead in usage, a key proxy for developer engagement and real-world application, suggests China's AI platforms are achieving significant product-market fit. The key question for investors is how, and when, this dominance in usage will translate into revenue for companies like Tencent (00700.HK). The high volume indicates a potential for significant future earnings from API calls and enterprise services, but also signals a highly competitive, and potentially lower-margin, domestic market.
The diverging usage numbers may also reflect different underlying strategies. The Faraday Future press release [1] draws a contrast between a "full-stack in-house development model" like Tesla's and an "open-source and open developer platform" approach. This mirrors the AI landscape, where some U.S. leaders focus on highly integrated, proprietary systems, while parts of the Chinese market are fostering a more open, ecosystem-driven approach that can lead to explosive, broad-based usage growth. This open model, while potentially less defensible, can capture a wider developer base faster.
While this AI battle is currently measured in token usage, the broader competition is already arriving on North American shores. Canada's recent decision to slash tariffs on a quota of Chinese-made electric vehicles [2] provides a template for how Chinese technology can gain a foothold in the continent. Nearly 400 Canadian dealers have already applied to sell brands like BYD and Chery, drawn by the prospect of offering affordable EVs. This regulated entry into the auto market demonstrates a potential pathway for other Chinese-developed technologies, including AI services, to challenge U.S. incumbents directly.
The surge in token usage for models like Hy3 is a leading indicator of China's growing competitiveness in a critical technology sector. For investors, the data presents a complex picture: while Tencent is clearly winning developer mindshare, the path to monetizing this usage remains a challenge. The stock, a component of the Hang Seng Index, may see increased investor interest based on the strong adoption metrics, but questions around the long-term profitability of China's hyper-competitive AI market will remain.
This article is for informational purposes only and does not constitute investment advice.