China's Wenzhun raises $45M Series A extension for data model
What's the deal? Wenzhun Intelligent TechnologyDealroom has a profile for this one. Try Dealroom →, a Chinese data foundation model startup, has raised a US$41.3M Series A extension. The round drew CDH Investments, Shanghai Guotou Xiandao, and Xuhui Science and Technology Venture Capital, among other backers.
The track record: The announcement marks Wenzhun's fourth consecutive round since its founding in October 2024. Earlier backers include the Xiong'an fund — whose US$2.75M angel cheque was its first ever — plus Oriental Fortune CapitalDealroom has a profile for this one. Try Dealroom →, Guoke Investment, and HM CapitalDealroom has a profile for this one. Try Dealroom →.
What's the endgame? Spun out of a Tsinghua University team led by chief scientist Cui Peng, Wenzhun is building what it calls a Large Data Model (LDM), aimed at structured industrial and scientific data rather than the text that large language models handle. Its latest model, LimiX-2, topped all entrants in three international structured-data benchmarks — TabArena, TALENT, and BCCO.
Why now? Wenzhun says it closed the round despite a cooling primary market. One top fund that missed the previous round locked in early this time, and a rarely active fund secured a large allocation, according to the company. Its valuation has multiplied several times over in two years.
Guoke Investment, which led the Series A alongside HM Capital, said it wants LimiX to become "one of the core foundation models for China's industrial and scientific fields" (translated from Chinese). HM Capital pointed to the team's decade of work in causal learning as what drew it in.
The signal: The round reflects a wave of capital chasing Tsinghua-linked research teams building physical-world and data AI. Comparable startups such as LiberAI and OriginFlow have reportedly raised over US$137.5M and US$68.8M respectively. Topping a benchmark and raising four rounds signals investor conviction in an early bet — not yet proof of commercial scale.
Read more: 36kr.com
Image credit: "Research On Massive Amount of Climate Data Requires Fast Access" by j_cadmus, licensed under CC BY 2.0