Ahead of its third anniversary, Infinigence announced that it has raised over 700 million yuan in new funding, reinforcing its position as one of China’s leading AI-native infrastructure companies by total financing scale. The round was co-led by Hangzhou Hi Tech Financial Investment Holding Group Co., Ltd. and Grand Mount Capital (GMC), with follow-on investments from Guoxing Investment, Chindata Group, GF Qianhe, Leaguerfund, China Insurance Investment, AEF NextGen, Teng Rui Investment, Colorlight, China Capital, and Will (Wizard Intelligence Learning Lab). Existing shareholders Legend Capital, Fortera Capital, and Yuanzhi Future also participated in this round.
The continued support from investors, along with Infinigence's rapid growth, underscores its consistent alignment with the key technological waves shaping the evolution of the AI industry. Over the past three years, the global AI landscape has moved through several defining phases, from the “hundred-model battle” to the rise of AI agents. More recently, the Agentic AI-driven token economy has entered a period of rapid expansion. Building on its strengths in software–hardware collaboration and multi-source heterogeneous, Infinigence has established a strong AI-native foundation. It has also been quick to anticipate the shift toward multi-agent systems and end-to-end deployment, positioning itself early in the development of the next generation of self-evolving Agentic Infra.
This funding round also underscores strong investor conviction in the foundational role of AGI infrastructure in the token economy era.As the global AI industry shifts from a “model race” to a focus on real-world AI productivity, Infinigence is seeing rapid growth driven by its differentiated Agentic Infra stack, built around high efficiency, large-scale systems, and intelligent orchestration. The company is increasingly positioning itself as a key hub in China’s emerging token economy.
The proceeds will be focused on three areas: first, strengthening its leadership in multi-source heterogeneous and expanding access to highly optimized compute capacity in the token economy era; second, further advancing software-hardware collaboration to improve production efficiency, from “electricity to tokens”; and third, developing agentic AI infrastructure, leveraging its enterprise agent platform to deliver solutions across industries and enhance the transformation from “tokens to productivity.”
The Token Economy Wave Has Arrived, Ushering in a New Era of “AI Productivity”
The global AI industry is undergoing a profound paradigm shift. As AI agents move from concept to large-scale deployment, “tokens”—the fundamental unit for measuring AI output—are emerging as a new core resource driving both the digital economy and the physical world.
As of the end of April this year, Infinigence’s Agentic MaaS platform recorded more than a 20x increase in average daily token usage compared to the end of last year, alongside continuous day-zero integration of open-source models.According to the latest data from China’s National Bureau of Statistics, average daily token usage in China had surpassed 140 trillion as of March, up more than 40% from the end of last year. AI is being rapidly embedded across industries, driving the continued expansion of the digital economy. This trajectory echoes the early days of mobile data in the 3G era, when monthly usage of 100MB was considered meaningful. Today, token consumption is still at an early stage of exponential growth.
The long-term value of AGI infrastructure lies not only in technological leadership but also in fostering a sustainable token economy ecosystem.
In March this year, Xia Lixue, co-founder and CEO of Infinigence, first introduced the concept of “Tokenomics with Chinese Characteristics” at the 2026 ZGC Forum. The framework envisions a complete economic loop—from input and energy consumption to tokens, productivity, and value—positioning tokens as more than a technical unit of measurement, but as a core economic variable underpinning the future of AI development.
With abundant energy resources, a complete AI industrial chain, and the world’s largest AI application market, China is well positioned to extend the “Made in China” model into the AI era. Through its high-efficiency, large-scale, and intelligent Agentic Infra, Infinigence aims to convert these structural advantages into AI-driven productivity at scale, contributing to global AI development from a China-led perspective.
In response to this structural opportunity, Infinigence has formally introduced its AI Productivity Formula, tightly linking the token economy with its core technical capabilities. The framework also offers a new lens for addressing a fundamental question: how AGI infrastructure can translate into productivity gains across the broader economy.
Introducing the AI Productivity Formula: Intelligent Scale × Token Production Efficiency × Token Value Conversion
Intelligent Scale: The scale of multi-source heterogeneous computing that can be optimized to its limits through advanced engineering. In the current phase of rapid token economy growth, compute scale has become a key constraint on the upper bound of intelligence. The supply and cost structure of compute from individual chip vendors is increasingly limiting large-scale AI deployment. By leveraging its multi-source heterogeneous, Infinigence enables cross-layer optimization across diverse models, algorithms, and hardware, significantly expanding usable compute capacity and easing today’s compute bottlenecks.
Token Production Efficiency: Producing more tokens with the same resources, measured in “tokens per second (tokens/s).” Through its Agentic Infra Platform, Infinigence has built a large-scale integrated model training and inference platform for AI-native enterprises. Through deep software-hardware collaboration and cross-stack design, it maximizes the effective utilization of underlying compute resources.
Token Value Conversion: Generating greater productivity from the same number of tokens, measured in “productivity per token.” The Infinigence platform integrates leading large language models, including Kimi, Zhipu, DeepSeek, Tongyi Qianwen, and MiniMax. It further enhances and unlocks the capabilities of trillion-parameter open-source models, delivering system-level token productivity services across industries such as AI-native enterprises, manufacturing, media and entertainment, and smart devices.
Multiplying “token production efficiency” by “token value conversion” yields “productivity per second.” Across industrial history—from the Industrial Revolution to the rise of the internet—major technological shifts have ultimately been about increasing society’s productivity per second. Today, tokens are emerging as a new unit of productive value, embedded in the ongoing expansion of large-scale productivity.
Xia Lixue, co-founder and CEO of Infinigence, stated: “In the token economy era, AGI infrastructure plays a role similar to a refinery in the petrochemical industry. It converts energy into digital oil (tokens), serving as the foundational layer for AI applications and services and acting as a key hub in the token economy. The AI Productivity Formula we introduced is our systematic answer to how to build Agentic Infra that is highly efficient, large-scale, and intelligent. The over 700 million yuan in new funding will help accelerate progress across these three dimensions, offering a China-driven approach to addressing the core challenges of the global token economy.”
Agentic MaaS Platform: Daily Token Usage Surge More Than 20-Fold
Infinigence’s Agentic MaaS platform is built on a multi-layer architecture, providing full-stack capabilities—from model integration and traffic management to inference optimization and resource scheduling—for LM providers and AIGC application developers. The platform also uses agent-based coordination to dynamically adjust underlying model resources based on system performance and load, ensuring 24/7 high availability.
It delivers high-performance optimization for leading domestic open-source models such as Zhipu, Kimi, DeepSeek, Tongyi Qianwen, and MiniMax, and has performed deep inference optimization for trillion-parameter LMs.
In real-world production environments, the Agentic MaaS platform demonstrates strong performance across multiple dimensions.In terms of execution accuracy, it supports complex toolchains while maintaining high consistency with original model outputs (accuracy alignment rate above 99.9%). In terms of efficiency, it improves throughput by 2–3x, reduces latency by 50%, and keeps first-token latency under 500 ms. It also achieves 99.95% enterprise-grade availability, providing stable support for production workloads.
Currently, the Agentic MaaS platform supports more than 160 LMs, all available out of the box.As of the end of April, the platform’s average daily token usage had increased more than 20-fold compared to the end of last year, forming a strong growth flywheel: lower inference costs → deeper collaboration with leading LLM providers → accumulation of real-world production data → continuous optimization of inference systems → improving performance and reliability → sustained growth in token usage.
Investor Comments
The head of Hangzhou Hi Tech Financial Investment Holding Group Co., Ltd., a co-lead investor, stated: “High-Tech Jintou Group is committed to empowering technological innovation with capital and serving the real economy. We focus on national strategic sectors such as AI, integrated circuits, and the digital economy, and aim to build core capabilities in indigenous innovation. Our investment in Infinigence reflects our recognition of the team’s Tsinghua background, strong technical expertise, and proven execution capability, as well as our long-term confidence in China’s AGI infrastructure development. With support from the High-Tech Zone’s industrial ecosystem, policy resources, and post-investment services, we will provide integrated support across capital, application scenarios, and ecosystem development. We look forward to helping Infinigence strengthen its position in AGI infrastructure and accelerate the commercialization and scaling of its technologies.”
Zhang Shanshan, President of Grand Mount Capital (GMC), a co-lead investor, stated: “AGI infrastructure is the foundational layer of the token economy, with an impact comparable to how electricity and the internet reshaped productivity in previous industrial eras. As a co-lead investor, GMC will leverage its AI ecosystem resources, including data centers and industry application scenarios, to deepen integration between capital and industry. We are committed to supporting Infinigence’s growth and helping it expand its role as a hub in the token economy. We believe Infinigence is well positioned to become a leading benchmark in China’s AGI infrastructure space.”
Ho Ting-hong, Head of the Alibaba Hong Kong Entrepreneurs NextGen Fund, stated: “As Infinigence’s first overseas investor, the AEF NextGen Fund is deeply honored to participate in this funding round. Under the outstanding leadership of its management team, Infinigence is playing a key role in driving the innovative development of the token economy in China and globally. With our resources and network in Hong Kong and international markets, we look forward to deepening our partnership with Infinigence and jointly exploring new opportunities in Hong Kong and international markets.”
