Back to feed
News Story
量子位
1 sources

Wuwen Xinqiong Raises 2.3 Billion Yuan in Series C, Total Funding Hits 4.3 Billion Yuan, Setting New Record for Domestic AI Infra

AI-native infrastructure provider Wuwen Xinqiong has completed a 2.3 billion yuan Series C round, bringing its total funding to 4.3 billion yuan, a new record for both single-round and cumulative funding in China's AI Infra sector. Investors include state-owned entities, industrial players like Lenovo, Tencent, and Baidu, and financial institutions such as Sequoia China. The company focuses on the middleware layer between models and chips, aggregating heterogeneous compute resources and optimizing software-hardware co-design, having adapted to 16 chip types and serving 37,000P of compute, against a backdrop of widening compute gaps driven by agent proliferation.

SynthePulse Insight · AI deep readingMembers

2.3 Billion Yuan Betting on "Not Building Data Centers": What Wuwen Xinqiong's Funding Record Is Really Wagering On Is the Compute Middle Layer

Version 1 · 1 source

Wuwen Xinqiong has closed a 2.3 billion yuan Series C, bringing cumulative funding to 4.3 billion yuan, setting new single-round and cumulative records for China's AI-native infrastructure track, and has reportedly filed a confidential IPO application with the Hong Kong Stock Exchange. It doesn't train models, doesn't build applications, doesn't build data centers—it only does the layer between models and chips, and this money is betting on whether that position can keep monetizing in the Agent era.

  • Wuwen Xinqiong closed a 2.3 billion yuan Series C, with cumulative funding of 4.3 billion yuan, both setting new records for China's AI-native infrastructure track.
  • The company doesn't train models, doesn't build applications, and doesn't own physical data centers; instead it weaves compute resources from data centers across regions into a network through software orchestration.
  • The platform claims to have adapted to 16 types of chips, serving 37,000P of compute, with new chip adaptation taking about 2 working days and domestic chip adaptation costs cut by over 70%.
Open section navigationA Record-Breaking Funding Round, and the Price It Didn't Disclose

A Record-Breaking Funding Round, and the Price It Didn't Disclose

Wuwen Xinqiong has completed a Series C funding round of 2.3 billion yuan; adding previous rounds, cumulative funding stands at 4.3 billion yuan, with both the single-round and cumulative amounts setting new records for China's AI-native infrastructure track. This is the most certain part of this event: the amount and round are directly given by the report and constitute confirmed facts.

But the other most critical number in the funding narrative is absent—the post-money valuation of this round was not disclosed in the source, and the specific list of investors for the 2.3 billion yuan and each investor's contribution amount were likewise not stated. This means outsiders cannot judge from public information how much equity this money bought, nor what the "record" means in valuation terms.

The investor structure itself was fully listed: state-owned institutions include Hangzhou High-Tech Jintou, Social Security Zhongguancun Fund (managed by Legend Capital), Zhuhai Technology Industry Group, Xiamen Guoxing Investment, Xuhui Capital, Futeng Capital, China Insurance Investment, and Guokai Venture Capital, among others; industrial investors include Dabanc Holding, Lenovo, Jiukun Venture Capital, Tongge Venture Capital, Tencent, and Baidu, among others; financial investment institutions include Sequoia China, Hongtai Fund, Huiyuan Capital, Qiming Venture Partners, Fortune Capital, Shunwei Capital, and Lisi Capital, among others. The simultaneous presence of state-owned, industrial, and financial capital in three categories is the most noteworthy structural feature of this funding round.

A further signal comes from overseas. Background data cited in the report shows that Fireworks AI, which does the same business, saw its valuation rise from $4 billion to $17.5 billion (up 337%), Baseten rose 160%, and Together AI rose 151%, with Nvidia investing in all three. It must be made clear that this is background data cited by the report and has no direct correspondence to Wuwen Xinqiong's own valuation; it cannot be used to infer its pricing.

Putting these points together, a bounded inference can be made: the fact that capital is willing to provide a track-record-level amount without a disclosed valuation suggests the pricing basis is more likely to come from technical metrics and positioning narrative rather than disclosed financial performance. But this inference cannot be confirmed by any number in the source.

Free for now

Read the full analysis

6 more sections of analysis, plus the full takeaway

Loading

Credibility boundary

All facts in this article come from a single source (a QbitAI report, source_tier 2). Among them, the 2.3 billion yuan Series C, cumulative 4.3 billion yuan, investor categories and list, and not owning physical data centers are confirmed facts; adaptation to 16 types of chips, 37,000P of compute, 2-working-day adaptation, adaptation cost cut by over 70%, 97.6% utilization in thousand-card heterogeneous mixed training, 70-billion-parameter heterogeneous mixed training, 165% performance improvement in 4,000-kilometer cross-domain mixed training, 41% and 68% improvements in multi-data-center and hybrid cloud mixed training, 41% to 75% higher training throughput for trillion-parameter MoE, PDD's 51.5% first-Token latency reduction and 37.5% per-Token cost reduction, and average daily Token usage growing more than 40-fold in half a year with over 95% of inference tasks from Agents are all self-reported by the company or the reporting party, lacking independent third-party verification, and are treated in this article as source claims. The Hong Kong Stock Exchange IPO application information comes from Bloomberg citing people familiar with the matter and is second-hand information not officially confirmed. The valuation increases of the three overseas AI infrastructure companies are background data cited by the report and have no direct correspondence to Wuwen Xinqiong's own valuation. This round's post-money valuation, specific investor contribution amounts, test conditions for technical metrics, and cooperation terms and revenue contributions with leading large model companies were all not disclosed in the source.

Primary report

量子位

Primary source