700 Million Yuan! The Birth of China’s Largest AGI Infrastructure Financing

On May 7, 2026, Wuwen Xinqiong, a leading player in AGI infrastructure, announced it has secured over 700 million yuan in financing. This figure not only sets a record for financing in China’s AI infrastructure sector but also brings renewed attention to the long-underestimated field of AI infrastructure.

In the past three years, capital has primarily chased model capabilities, but this logic is shifting: the infrastructure supporting the large-scale deployment of models is becoming a new value anchor.

Wuwen Xinqiong is a typical example of this trend. Founded in 2023 and emerging from the Tsinghua University system, the company focuses on transforming computing power into high-quality tokens. It currently serves several leading model companies, including Yuezhi Anmian and Zhiyu, and has built a technological system characterized by “diverse heterogeneous + software-hardware synergy + autonomous AI.”

Recently, at a roundtable during the Beijing Zhongguancun Forum, Wuwen Xinqiong’s co-founder and CEO, Xia Lixue, shared impressive data: since the end of January 2026, the average token call volume on its platform has doubled every two weeks. He remarked, “The last time I saw such a steep growth curve was during the explosion of mobile data in the 3G era.”

The financing round for Wuwen Xinqiong features an “all-star” lineup of investors, led by Hangzhou High-tech Jin Investment Group and Huiyuan Capital, with participation from Guoxing Capital, Qinhuai Data, Guangfa Qianhe, Lihua Qingtong, Zhongbao Investment, AEF NextGen, Tengrui Capital, Kalete, CITIC Capital, and Kuan De Intelligent Learning Laboratory, along with additional investments from existing shareholders Junlian Capital, Shanghai Guotou Futeng, and Yuanzhi Future.

This diverse financing consortium sends a strong signal: recognition of AGI infrastructure has evolved from a consensus within the tech community to a consensus across the industrial ecosystem.

Why Independent Infrastructure is Growing Amidst Major Players

A common viewpoint in the market is that as major companies like Alibaba and ByteDance build their own computing power and model systems, the space for independent AGI infrastructure providers is shrinking. However, the reality is quite the opposite. The AGI infrastructure developed by major companies is essentially “self-use infrastructure” tailored for specific business ecosystems, with highly customized architectures, scheduling strategies, and even chip selections.

In contrast, truly public AI infrastructure capable of serving diverse models, chips, and scenarios is becoming increasingly scarce. This explains why independent firms like Wuwen Xinqiong are experiencing a surge in demand.

The data reflects this vividly. At the time of announcing its 700 million yuan financing, the company disclosed that as of the end of April this year, the average daily token call volume on its Agentic MaaS large model service platform had increased over 20 times compared to the end of last year. On a broader industry level, China’s average daily token call volume has surpassed 140 trillion. This growth curve closely mirrors the explosive path of mobile data during the 3G era.

With applications like OpenClaw reshaping demand structures, complex agent tasks—such as automatically analyzing a hundred-page financial report or planning an international trip and completing all bookings—can consume tokens in the tens or even hundreds of thousands per task. Infrastructure is no longer merely supporting model operations; it directly determines the upper limit of AI productivity.

In this context, the value of independent third-party infrastructure has been amplified like never before. As Xia Lixue stated, “Traditional infrastructure is designed for human engineers and programmers, where task initiation takes minutes or even hours. Agents can initiate tasks in milliseconds, and existing systems simply cannot adapt.” This structural mismatch leaves significant value space for independent infrastructure providers that can be redesigned for the agent era.

Redefining Infrastructure Value: From Computing Power to AI Productivity Formula

At first glance, many may mistakenly believe that the core competitiveness of AI infrastructure companies lies in “computing power scale.” This is a profound misunderstanding and one of the industry’s long-standing pain points.

Historically, the quality of infrastructure has been measured by simple and crude standards: how many cards? How fast? This is akin to evaluating a factory based solely on its “power generation” while completely ignoring whether the products produced can be sold at a good price.

Now, a more fundamental question is emerging: How much of the tokens produced by computing power are converted into real value?

Wuwen Xinqiong has clearly dissected this value chain: input → electricity → tokens → productivity → value. In this chain, traditional infrastructure focuses only on the first two links—how to produce more tokens with less electricity and faster speeds. However, the real bottleneck often lies in the latter two links: How much productivity can these tokens drive? How much commercial value can this productivity ultimately convert into?

Based on a deep understanding of this pain point, Wuwen Xinqiong has proposed its own solution framework—the AI Productivity Formula:

AI Productivity = Intelligent Scale × Token Production Efficiency × Token Value Conversion

This formula’s three factors redefine the value connotation of AGI infrastructure.

Factor 1: Intelligent Scale, which depends on the scale of diverse heterogeneous computing power that can be optimally enhanced through technology. The supply and cost structure of a single chip manufacturer are currently restricting the process of AI large-scale deployment. Wuwen Xinqiong utilizes a core technology of “diverse heterogeneity” to achieve cross-layer extreme optimization between various model algorithms and chip hardware, significantly enhancing the usable scale of computing power and solving the current computing power shortage dilemma.

Factor 2: Token Production Efficiency is Wuwen Xinqiong’s signature strength. The company has launched a large-scale model training and inference integrated platform for AI-native enterprises through its Agentic Infra autonomous AI base, optimizing through software-hardware synergy, and deeply optimizing and jointly designing across software and hardware to maximize chip application computing power.

Factor 3: Token Value Conversion is the more critical leap. A high-quality token, if used for precise prompts or complex task breakdowns, can generate value hundreds or thousands of times more than low-quality tokens. Wuwen Xinqiong’s platform has already connected with leading large models such as Kimi, Zhiyu, DeepSeek, Tongyi Qianwen, and MiniMax, further optimizing and leveraging the intelligent upper limit of trillion-parameter open-source large models, and providing system-level token productivity services to AI-native enterprises, manufacturing, cultural and film industries, and smart terminals.

These three factors together form a closed loop: from energy to tokens, then to productivity and economic value; marking the industry’s competition has shifted from simple card-stacking contests to a new stage of calculating economic accounts.

This also explains why Wuwen Xinqiong can simultaneously serve leading model companies like Yuezhi Anmian and Zhiyu. When infrastructure becomes a neutral and efficient productivity tool, it often gains the trust of all players due to its pure enabling role.

Who is Betting? Insights into the “All-Star” Investor Lineup Behind the Industrial Ambition

It is precisely based on the redefinition of infrastructure value that capital has recognized the long-term value of this sector.

Wuwen Xinqiong’s recent financing of over 700 million yuan features an “all-star” lineup of investors, each bringing clear industrial logic to the table.

First, there is the entry of government industrial capital. Local funds represented by Hangzhou High-tech Jin Investment and Huiyuan Capital view AI infrastructure as the “high-speed rail tracks of the digital economy era.” From the perspective of government-side capital, tokens are akin to electricity and data, serving as basic production factors, while infrastructure determines the efficiency of factor circulation.

Next, the backing of national-level capital is significant. The entry of long-term funds like Zhongbao Investment indicates that the market is beginning to assess AGI infrastructure from the perspective of infrastructure rather than technology projects. This type of capital values not short-term explosions but the long-term compound returns brought by sustainable tokens.

Additionally, there is the “vote with feet” approach from industrial capital. The participation of companies like Qinhuai Data and Kalete indicates that infrastructure capabilities are beginning to deeply bind with the real industry. For data center operators, efficient token production capabilities mean higher computing power utilization; for manufacturing enterprises, it means the true activation of data and AI capabilities.

Looking globally, the logic behind this all-star lineup aligns closely with the latest trends in international capital markets.

Overseas, the new computing power cloud sector is rapidly rising. In the North American AI infrastructure sector, CoreWeave has seen its market value soar after going public, becoming a crucial part of AI computing power supply; Baseten’s valuation has multiplied within a year, attracting investments from industry giants.

Global capital has reached a consensus: Computing power itself is not valuable; the ability to efficiently convert computing power into tokens is what holds value. As Wuwen Xinqiong’s co-founder and CEO Xia Lixue metaphorically puts it, it resembles a “refinery” in the petrochemical industry, transforming raw energy into high-value-added “digital oil”—tokens.

Conclusion: Evolution from World Factory to World Token Factory

The previous round of China’s economic growth engine transformed manufacturing cost advantages into exports of “Made in China” goods. In the new wave of AI, China is establishing a high-efficiency “token factory” by leveraging its energy structure advantages, a complete manufacturing system, and strong engineering capabilities.

Wuwen Xinqiong’s successful financing sends an industrial signal: China is evolving from the “world factory” to the “world’s token factory.” By providing high-quality, stable, and sustainable token services, China has the capability to contribute a unique and indispensable “Chinese solution” to global AI development.