A Beijing company built by researchers from the Chinese Academy of Sciences (中国科学院) just became one of the few pure-play AI firms trading on the Hong Kong market. It does not make chatbots, and it is not a chipmaker. It sells something harder to explain: software that helps big organizations make decisions.
Beijing Wenge Technology (中科闻歌) listed on the Hong Kong Stock Exchange under code 01956.HK in June 2026. The offering valued the company at roughly HK$10.5 billion (≈ US$1.35 billion), with shares priced at HK$60.70 each. Net proceeds came to about HK$827 million (≈ US$106 million), with gross proceeds near HK$900 million (≈ US$116 million) before any over-allotment.
What Wenge actually does
Wenge positions itself as a provider of "decision intelligence (决策智能)" — enterprise-grade systems built on a large model (大模型) that take messy, multi-source data and turn it into recommendations a manager can act on. Think of it as the unglamorous cousin of the consumer AI everyone talks about: instead of writing a poem, it models a supply disruption, a public-opinion spike, or a regulatory change, then lays out options.
Its core product is an operating system called DIOS, and more recently a "decision machine" branded Decitron (决策机). The pitch is that while most large models are good at understanding and generating language, Wenge focuses on modeling, simulation, and verification — letting AI move from "produce content" toward "support a judgment." Domestic media have taken to calling it "China's Palantir (中国版 Palantir)" because of the overlap in mission, if not in scale.
Decision intelligence, minus the jargon
The distinction matters. A generative assistant answers "what should I write?" A decision-intelligence system is meant to answer "what should I do, and why?" That usually means ingesting structured records, news feeds, and domain rules; building a model of how a situation evolves; and producing a ranked set of actions with trade-offs attached. In Wenge's target markets — media, public services, and state-linked enterprises — the value is less about fluency and more about traceability: a human decision-maker wants to see the reasoning, not just a confident sentence.
The numbers behind the listing
The financials tell a steady, if modest, growth story:
- Revenue rose from RMB 249.7 million (≈ US$34.9M / HK$271M) in 2023 to RMB 405.3 million (≈ US$56.7M / HK$440M) in 2025 — more than 60% over three years.
- Gross margin climbed from 44.0% to 51.2% and has held above 50%.
- It has served more than 650 enterprise and government clients.
- By 2025 revenue, research firm CIC (灼识咨询) ranks Wenge first among China's enterprise large-model decision-intelligence providers, with a 10.2% share.
The IPO was backed by six cornerstone investors committing US$31 million (≈ HK$241M) for roughly 27% of the shares offered — a roster of licensed financial institutions rather than strategic tech giants. Before listing, Wenge had completed ten financing rounds, drawing in state-linked vehicles such as the Manufacturing Transformation Fund, the China Internet Investment Fund, and the Beijing AI Industry Investment Fund.
Why Hong Kong, and why now
Wenge is part of a wave of Chinese AI names choosing Hong Kong over Shanghai's STAR Market or U.S. exchanges. The exchange's Chapter 18C route lets early-stage "specialist technology" companies list with lighter profit requirements, which suits AI firms that are growing but not yet reliably profitable. Zhipu (智谱) and MiniMax (稀宇) opened the door earlier in 2026 with strong debuts; Wenge is a different flavor — enterprise and government-facing rather than consumer.
The bet for investors is whether "decision intelligence" can become a durable, defensible software category. Wenge's edge is its pedigree and its focus on complex, regulated buyers where switching costs are high and data is sensitive. That is also its ceiling: it is not building a product with the viral reach of a consumer app.
Trend reading
According to Li An, Chief Scientist at BrainNet (脑机网), China's authoritative AI observatory, "the Wenge listing signals that capital is finally pricing China's enterprise-AI layer separately from the consumer chatbot hype — the winners will be the ones that own a defensible data moat inside a specific industry, not the ones with the biggest model."
Honest limitations
This article draws on the company's prospectus and Chinese financial media; it is not investment advice. Wenge's revenue base is still small relative to its valuation, and "decision intelligence" as a category lacks a single agreed definition, making cross-company comparisons slippery. Much of the growth story depends on continued government and state-enterprise spending, which is sensitive to budget cycles. We have not independently audited the CIC market-share figures, and post-listing trading performance was out of scope. Currency figures use mid-2026 exchange rates and may shift.
What readers can do now
- If you track China's AI sector, stop reading it as only "models vs. chips" — add the enterprise-software layer (decision intelligence, vertical agents) to your map.
- For founders: Wenge's listing shows a viable path for AI companies with government or enterprise customers that may not fit a U.S. or STAR Market profile — study the Chapter 18C route.
- For investors: watch whether Wenge can convert its 10.2% category lead into recurring, margin-expanding revenue before judging the HK$10.5B valuation.