Token Demand Surges 1000-Fold, 2.2 Billion Yuan Pours Into AGI Infra Leader

At this year's GTC conference, NVIDIA CEO Jensen Huang declared that the company is no longer just a chip or GPU company but has fully transformed into a full-stack AI infrastructure company. This announcement once again thrust "AI infrastructure" into the industry spotlight. However, Infinigence AI, a leading Chinese AGI infrastructure player, had already rooted itself in AI's underlying infrastructure long before the trend took off. Over the waves of industry evolution, it has continuously adapted and become a benchmark for China's AI infrastructure development.

Today, its token production and other services have permeated the underlying layers of major models such as Kimi, GLM, MiniMax, and DeepSeek — almost every user of Chinese large models has indirectly used its services. Recently, the company announced that it had secured an additional over 700 million yuan in funding. This scale firmly places it in the first tier of domestic AI-native infrastructure enterprises, making it one of the fastest-growing newcomers in China's underlying AI track.

Public data shows that as of the end of April 2026, Infinigence AI's MaaS platform daily token call volume had increased more than 20-fold compared to the end of 2025, a growth rate dozens of times the national average. Since late January, token call volume has doubled every two weeks. In just three years since its founding, Infinigence AI has secured nearly 2.2 billion yuan in total funding from top-tier investors, becoming an indispensable partner for many leading domestic large models.

The driving force behind this growth is the token economy wave now sweeping the AI industry. The Agent era has brought three fundamental disruptions to infrastructure. First, Agents have transformed large models from "chatbots" into "digital workers" capable of autonomous planning, execution, and review. Single-task token consumption has skyrocketed from hundreds to millions, straining computing resources. Second, interaction speed has shifted from minutes to milliseconds, exposing latency bottlenecks in traditional architectures. Third, stability requirements have become far more stringent, as Agents run continuously for hours, demanding high reliability and fault tolerance.

National Bureau of Statistics data shows China's total daily token calls have exceeded 140 trillion, a year-on-year increase of over 40%. Infinigence AI's growth is even more remarkable. Its CEO Xia Lixue noted that the growth curve reminds him of the explosion of mobile data traffic during the 3G era, which underpinned a golden decade for mobile internet. Now, the exponential rise in token calls signals that AI is moving beyond the concept stage toward widespread adoption.

Why does the industry need an independent third-party MaaS platform? Just as Apple, Qualcomm, and AMD rely on TSMC, all model companies need efficient inference computing and a trustworthy independent platform. Infinigence AI's differentiation lies in its neutrality and full commitment to infrastructure. Its Agentic MaaS platform provides high-performance optimization for popular open-source models like GLM, Kimi, MiniMax, DeepSeek, and Tongyi Qianwen, with accuracy alignment over 99.9%, throughput improved 2-3 times, latency reduced by 50%, and first-token latency under 500ms. Enterprise-level high availability reaches 99.95%.

To meet Agent-era demands, Infinigence AI has completed a comprehensive upgrade of its Agentic Infra technology system, addressing high latency, poor stability, and insufficient concurrency. It now supports over 160 models available out-of-the-box, with rapid adaptation for new open-source models. The company's technical prowess has earned industry recognition: in February 2025, MIT Technology Review highlighted Infinigence AI as one of four Chinese AI startups worth watching alongside DeepSeek, Zhipu AI, and others.

In March 2026, at the ZGC Forum, Xia Lixue shared the stage with Kimi founder Yang Zhilin and Zhipu CEO Zhang Peng, revealing how Infinigence AI serves as the infrastructure and token service provider for these leading models. Its partnership network also includes DeepSeek, MiniMax, and Tongyi Qianwen, covering most top domestic models. If these models are power plants generating AI capabilities, Infinigence AI is the super grid connecting them, integrating distributed computing resources and delivering token production and scheduling to industries and devices.

Following the latest funding round, Infinigence AI unveiled a new AI productivity formula: AI Productivity = Intelligence Scale × Token Production Efficiency × Token Value Conversion. This formula elevates tokens from a mere technical metric to a core economic variable driving AI industrialization. The company's long-term goal is to build high-efficiency token factories in China, leveraging domestic energy and industrial advantages to produce standardized, high-quality token products for global developers and enterprises.

The market potential of the token infrastructure track is reflected in investor enthusiasm. The recent IPO of CoreWeave, a US-based AI computing service provider, saw its market cap soar from $23 billion to $66 billion, a 189% increase. Infinigence AI's latest funding round attracted a cross-sector investor base including government capital, financial institutions, and manufacturing companies, indicating broad recognition of the token economy's value beyond the tech sphere.

In summary, Infinigence AI occupies a golden position in China's AI industry, serving as the "water and electricity" for the entire ecosystem. As token demand continues its explosive growth, the company is poised to be a key enabler of China's AI industrialization and global competitiveness.