(Yicai) Oct. 14 -- Chinese artificial intelligence models are gaining ground overseas as more companies turn to open-source alternatives, potentially shifting market share and future revenue away from US closed-source rivals, according to an AI investor.

Although Chinese LLM developers are unlikely to match the revenue scale of US giants such as OpenAI and Anthropic in the near term, their sales are growing rapidly, Cynthia Zhang, founding partner of FutureX Capital, told Yicai on the sidelines of the FutureChina Global Forum 2026, organized by Business China in Singapore. 

Despite trailing US rivals in revenue, Chinese models are leading in usage on some platforms. Chinese AI models have outpaced US models in weekly usage on OpenRouter for 24 consecutive weeks, according to the platform's latest data for the week ended Oct. 11. 

As Chinese open-source models narrow the performance gap with leading closed-source alternatives, major AI application developers are increasingly turning to them to build their own systems, Zhang said. This could reshape the competitive landscape for AI providers.

“Closed-source models may see their revenue decline, while open-source models’ revenue will move upward,” the founder of the Hong Kong-based venture capital firm said.

Among those embracing the technology are legal AI startup Harvey and multi-model agent platform Genspark, which are developing proprietary models based on Chinese open-source technology to reduce their dependence on overseas closed-source providers, according to Zhang.

“This trend has just begun, but it is growing very fast,” she said, adding that the shift reflects a broader push by major companies to take control of their AI capabilities. “The larger the company, the more it values its own AI capabilities, and the less willing it is to build its AI on someone else’s ecosystem or model.” 

Open-source models allow companies to customize their systems using proprietary data, giving them greater flexibility, Zhang explained.

Competition among LLM providers is also intensifying on the pricing front. According to Zhang, a leading US AI model typically enjoys high profitability for only about six months before prices begin to fall. This could erode the pricing power and profitability of US model developers as cheaper alternatives gain ground.

Overseas Chinese Teams Drive Global Expansion

Beyond technological advances and competitive pricing, overseas expansion also depends on reaching international customers.

Overseas Chinese-led technology teams are helping accelerate the global adoption of Chinese AI models by combining knowledge of China's technological strengths with insights into international markets, Zhang said.

One example is California-based AI infrastructure startup Fireworks AI, whose core management team is largely of Chinese descent, according to Zhang. The company deploys Chinese open-source LLMs for major overseas enterprises and has surpassed USD1 billion in annualized revenue, she said.

“People used to think that Chinese LLMs could only compete at the application layer, or only on price, but that is not the case. With strong engineering execution and commercial acumen, Chinese teams are taking leading positions in key parts of the global AI industry chain,” Zhang said.

But technological strength and commercial success are not enough to compete globally. Chinese LLM developers must also navigate increasingly complex regulatory requirements.

Data compliance has become a major obstacle to overseas expansion, Gu Zhaoqin, a partner at the Hongqiao branch of Duan & Duan Law Firm, told Yicai. The challenges stem from genuine data security and privacy concerns among foreign regulators and consumers, as well as geopolitically driven trade protectionism, Gu said.

“Chinese AI going global is not just about competing on technical capabilities; it also requires the ability to earn global trust through rules, transparency and security compliance,” the legal expert concluded.

Editors: Tang Shihua, Emmi Laine