A zone in Beijing has launched a program asking several banks to extend billions of yuan in credit to multiple companies across the AI industry chain, partly based on their token usage.

Termed “token loans,” the financing scheme was recently implemented by the Beijing Economic and Technological Development Zone, also known as Beijing E-Town, in the city’s southeast, and involves extending nearly 2 billion yuan ($300 million) in credit to several AI companies, domestic media reported Monday.

A token is the smallest unit of data an AI large language model can process. In March, the National Data Administration officially adopted “ciyuan” — combining “word” and “currency” — as the Chinese term for “token.” The move was largely seen as putting an economic spin on AI-led computing power.

In June of this year, China’s daily token consumption already exceeded 500 trillion, according to data from the Ministry of Industry and Information Technology. A Saturday report by the China Telecom Research Institute estimates that that number could rise to 100 quadrillion, or 200 times that figure by year-end.

Six domestic banks, including the Agricultural Bank of China, China CITIC Bank, Industrial Bank, Bank of Beijing, China Minsheng Bank, and Huaxia Bank, provided the loans. The credit is intended for AI companies with “high research investment and high technological barriers,” as well as growing computing needs, but without the physical assets or profits typically required to obtain bank financing.

In addition to token consumption, the banks also awarded the loans based on technological barriers, team strength, and companies’ business development prospects.

Guoqi Zhikong, a developer of automobile-focused connected cloud platforms and autonomous driving vehicles, received a 30-million-yuan credit line from China CITIC Bank. Meanwhile, Shenzhou Guangda, which provides computing infrastructure and enterprise-level AI agent solutions, received the same credit amount from Industrial Bank. 

The rapid growth of the AI industry has created a dilemma for tech startups, which often struggle to obtain traditional loans, an anonymous staff member from Beijing E-Town’s business and finance bureau told domestic media. Conventional lending models rely heavily on collateral and corporate profit statements, so even companies with strong technological capabilities often lack physical assets to pledge. In contrast, their opportunities for technological development and business expansion are constrained by profits that have yet to fully materialize.

In August, Bank of China’s branch in the southern city of Guangzhou launched the country’s first token loan scheme. Aimed at small- and medium-sized computing companies, it covers token supply, applications, and related services, with successful applicants eligible for up to 30 million yuan in credit.

Looking ahead, Lou Feipeng, a researcher at the Postal Savings Bank of China, said that before such loans can be adopted more broadly, the country must address several major issues, including exaggerated token-use data, inaccurate assessments of computing power consumption, and overestimation of borrowers’ repayment ability.

Editor: Marianne Gunnarsson.

(Header image: VCG)