China's Approaching.AI raises over 1.3 billion yuan to sell AI 'tokens'

Approaching.AI, a Beijing start-up whose Chinese name is Qujing Technology, has closed a Series B funding round of more than 1.3 billion yuan, roughly $180 million to $195 million depending on the exchange rate used, the company announced on Monday, October 26, 2026. It said the round takes its total fundraising in 2026 to 2.4 billion yuan. The company sells AI "tokens", the units of text that AI models read and write, through a cloud platform, and says it will use the money to expand capacity, upgrade the platform and invest in research on inference, the stage when a trained AI model answers users' requests.

Most attention in artificial intelligence goes to the companies that train giant models. But once a model is trained, every question a user asks has to be processed, and that work is called inference. Companies like Approaching.AI do not build the headline models themselves; they run models efficiently on large banks of chips and sell the output by volume, measured in tokens. A token is roughly a word or part of a word, so a long chatbot answer can use hundreds of them.

In China, this business has a particular twist. United States export controls limit Chinese companies' access to the most advanced American AI chips. Reports describe Approaching.AI's operations as "token factories" built on domestic Chinese chips. If it can deliver cheap, reliable inference on locally made hardware, it helps Chinese developers keep building AI products despite the restrictions. State-linked funds and a bank's investment arm are among its backers.

The size of the round is also a sign of where investors see money in AI. Selling inference is a volume business that grows with every new app, from customer-service bots to coding assistants. Investors are betting that demand for tokens will keep rising as AI moves from experiments to everyday use.

For India, the story is relevant to its own AI plans. India's government has been building shared computing capacity for start-ups and researchers under the IndiaAI Mission,. The Chinese example shows how inference platforms can become large businesses in their own right, and how countries try to build AI capacity on hardware they can control.

Short-term: The money lets Approaching.AI add computing capacity and compete for more customers among Chinese AI developers.

Long-term: If inference on domestic chips proves cheap and reliable, it could reduce Chinese AI companies' dependence on foreign hardware and intensify price competition in AI services.

Who is affected: Chinese AI app developers who buy computing power, domestic chipmakers, investors in AI infrastructure, and policymakers in the US, India and elsewhere watching how export controls shape the AI industry.

Questions and Answers