Zhipu’s Stock Soars After Chinese AI Startup's Annual Revenue More Than Doubles
Liu Jia
DATE:  Apr 01 2026
/ SOURCE:  Yicai
Zhipu’s Stock Soars After Chinese AI Startup's Annual Revenue More Than Doubles Zhipu’s Stock Soars After Chinese AI Startup's Annual Revenue More Than Doubles

(Yicai) April 1 -- Shares of Knowledge Atlas Technology Joint Stock, better known as Zhipu AI, surged after the Chinese artificial intelligence developer said it more than doubled revenue last year thanks to robust market demand for AI.

Zhipu [HKG: 2513], the first of China’s six “AI tigers" to go public, closed 32 percent higher at HKD915 (USD117) a share in Hong Kong today. The stock, which began trading on Jan. 8, has skyrocketed almost eight times in value from an initial public offering price of HKD116.20.

Revenue climbed 132 percent to CNY724 million (USD99 million) in the 12 months ended Dec. 31, the Beijing-based firm said in a financial report released yesterday. On-premises deployment accounted for 74 percent of the total.

Zhipu's net loss widened 60 percent to CNY4.7 billion (USD646 million) on the back of a 45 percent jump in research and development spending to CNY3.2 billion. Gross profit jumped 69 percent to CNY297 million, but gross profit margin shrank to 41 percent from 56 percent, crimped by an increased share of lower-margin cloud deployment business.

Annualized recurring revenue at Zhipu's open platform and application programming interface business reached about CNY1.7 billion, a 60-fold surge, Chief Executive Zhang Peng said on its earnings call. Engineering optimizations on the inference side lifted the gross margin of its Model-as-a-Service platform to 19 percent from 3.3 percent, above the industry average, he added.

Zhipu will continue along the path of being “China's Anthropic,” Zhang said, with API revenue as the starting point of a structural, long-term trend. He noted that the industry needs a simple and economical business model.

Zhipu's API pricing has surged 83 percent since the end of last year, yet demand still outstrips supply, Zhang said. Customers continue to accept the price increases in a competitive market crowded with compute-based API providers, which validates its leading model capabilities, he pointed out.

Income from enterprise general-purpose models was CNY366 million, while that from the open platform and application programming interface business almost quadrupled to CNY190 million. Revenue from the enterprise AI agent segment soared 249 percent to CNY166 million.

Deep integration

Nine of China's top 10 internet firms are deeply integrated with Zhipu's GLM models, the company said, adding that the platform had more than four million registered corporate and developer users as of this month, serving customers in over 218 countries and regions.

Independent large language model developers hold an inherent advantage in a period of rapid technological iteration, as large tech companies, constrained by resources, cannot remain competitive in every scenario and will continue to source technology from best-in-class suppliers, Zhang noted.

Regarding whether OpenClaw-driven price increases across the industry represent a lasting trend or a cyclical boom, Chairman Liu Debing said volume-price growth is sustainable, driven by improved model capabilities and shifting user demand. The company's API prices soared, but token consumption and user growth remained fast, he added.

Future token pricing will become tiered: low-complexity, lightweight tasks will tend toward low-cost or even free, ad-supported models, while high-complexity, high-reliability tokens will command sustained pricing power, with customers focused on the return on investment rather than unit price, according to Liu.

Over the next two to three years, the industry's core growth driver will likely be the scaled rollout of standardized API capabilities, with cloud LLM costs continuing to fall and AI compute increasingly concentrated in the cloud, he said.

Editor: Martin Kadiev

Follow Yicai Global on
Keywords:   Zhipu AI,annual results,net loss,API revenue,token economy,MaaS,ARR,large language model,AI monetization,Hong Kong listing,Z.ai