Under the promotion of the Chinese Communist Party in developing artificial intelligence (AI), many enterprises in mainland China are witnessing an increasing demand for AI. Over a dozen major banks have rolled out “Token Loans” as a loan product, yet experts believe three risks exist.
In the field of artificial intelligence and computing, Token is the smallest basic unit used when processing text and information by models. In March 2026, the Chinese National Data Bureau officially named Token in Chinese as “詞元”. At the 2026 China International Information and Communication Exhibition held on September 23, Academician Wu Hequan of the Chinese Academy of Engineering stated that as of August, the daily Token call volume in China has exceeded 5 quadrillion, with a compound annual growth rate of up to 333%.
According to a report by “The Paper” on October 7, after the call volume of Large Language Models (LLM) surged over a thousand times in two years, over a dozen banks in China seized the opportunity to introduce “Token Loans,” disrupting the traditional credit granting model. The first special financial product focusing on Token economy in Guangdong Province, named “Token Loan,” was officially launched in Guangzhou. China Bank, CITIC Bank, and Guangzhou Bank simultaneously introduced related products. The loan product, led by state-owned banks, will gradually extend to joint-stock banks and key regional city commercial banks.
Among them, the “Bank of China · Computing Power Token Loan” product released by Bank of China’s Guangzhou Branch is designed to offer three sub-products tailored to the supply, application, and service of computing power industries. With credit based on factors such as enterprise computing power Token output consumption, computing power service contract value, accounts receivable from computing power business, and Token commission settlement volume, companies can receive a credit total of up to 30 million yuan with a maximum term of three years.
In the Yizhuang Economic and Technological Development Zone in Beijing, Agricultural Bank of China, CITIC Bank, Minsheng Bank, Huaxia Bank, Industrial Bank, and Beijing Bank jointly provide credit support for the “Yi Enterprise Token Loan” product.
The report indicates that “Token Loans” target enterprises involved in high-tech industries such as computing power, large models, and artificial intelligence. Technical capabilities of enterprises have become an essential factor in the credit assessment by banks. Additionally, operating data such as order books, bank statements, and tax declarations have been frequently mentioned.
Regarding the scale of loan disbursement, the actual disbursement of “Token Loans” is currently far below the credit amount granted.
Tian Lihui, a finance professor at Nankai University, warned of three risks associated with the introduction of “Token Loans.” Firstly, there is currently no unified standard for verifying the authenticity of Token consumption data, posing a risk of fraudulent loans. Secondly, the rapid technological iteration in the AI industry means that today’s high consumption levels by companies do not guarantee continuous operating capabilities a year later, highlighting a risk of maturity mismatch. Thirdly, if local government interest subsidy policies are phased out too quickly, it may lead to bad loan exposure.
Wang Peng, Chief Analyst of Bote Consulting in the financial industry, suggested that banks should regularly reassess credit, adjust quotas based on real-time computing power needs, reduce credit lines when necessary, and increase collateral to avoid credit risks caused by the contraction of business activities from enterprises.
Credit risk has become one of the major risks in the banking industry. A report by Bloomberg on May 12, 2026, estimated that the real bad debt ratio of the Chinese banking industry is about 10%, corresponding to around $3 trillion in loans that should be classified as overdue but have not been accounted for.
In recent years, as the Chinese economy has been declining, financial risks in the banking industry have surged. According to data from the Chinese National Financial Supervision and Administration Bureau, more than 670 banks were closed in 2025, a record number. International rating agency Fitch released a report stating that after closing these banks, the number of physical banks in China decreased by 23% over four years.