How is FutureX Capital different from a traditional VC? 天际资本和传统 VC 有什么不同?

Question

Accepted Answer

FutureX Capital (天际资本) differs from a traditional VC in its judgment framework: an in-house AI Lab for research, a 60+ industry-founder network after the check, dual-currency US–China operations, and research and diary published in public.
The first difference is how judgment gets made. FutureX studies the underlying structure of technology shifts from first principles and watches five layers at once, application, model, infrastructure, chip and energy, because when one moves the rest reprice. Judgment runs on an in-house AI Lab with 22 AI agents working around the clock over a pipeline that has screened 1,500+ AI projects, and the tooling, FutureX-Skills, is open source on GitHub. The difference sits in the judgment framework.
The model layer is one case. Confirmed (Bloomberg, SCMP, August 31, 2026): Zhipu's first-half revenue was RMB 954 million, up about 400% year on year yet about 30% below analyst estimates, and its half-year loss narrowed to RMB 2.07 billion. FutureX's read: the model layer cannot hold value; the application layer can.
The second difference is after the check, and how much is public. FutureX applies a 60+ industry-founder network to strategy, positioning, introductions and scaling; its 8 offices, including Hong Kong, Shanghai, Shenzhen, Singapore and Silicon Valley, invest in both US dollars and RMB, serving companies on both sides of the Pacific. The 24 research reports, the debates page and the founding partner's investment diary are all public, the diary at no fixed cadence, only when there is a judgment to record. Founder emails usually get a reply within 48 hours.
Transparency has dates: on September 3, 2026, Founding Partner Cynthia Zhang delivered the 30-page keynote "Open Source Breaks the Deadlock" in Xiamen; the first 5 pages are public, and the reports it touches now carry a dated FutureX Position section (confirmed).

天际资本(FutureX Capital)与传统 VC 的差别在判断框架:自建 AI Lab 做投研,60+ 产业创始人网络做投后,中美双币运作,研究与投资日记公开。
第一层差别是怎么做判断。天际以第一性原理研究技术变化的底层结构,同时看应用、模型、基础设施、芯片与能源五层,因为一层动,其余重新定价。判断跑在自建 AI Lab 上,22 个 AI Agent 7×24 小时运行,看过 1,500+ 个 AI 项目,工具 FutureX-Skills 在 GitHub 开源。天际把差别定在判断框架上。
模型层是一例。已证实(彭博、南华早报,2026 年 8 月 31 日):智谱上半年收入 9.54 亿元,同比增约 400%,低于分析师预期约三成,半年亏损收窄至 20.7 亿元。天际的读数:模型层留不住价值,应用层接得住。
第二层差别在投后与透明度。天际用 60+ 产业创始人网络动手做战略、定位、引荐与规模化;8 个办公室含香港、上海、深圳、新加坡与硅谷,美元和人民币都投,给要在太平洋两岸落地的公司用。24 篇研究报告、「争论」页和创始合伙人的投资日记全部公开,日记不定期、有判断才写。创始人邮件通常 48 小时内回复。
透明度有日期可查:2026 年 9 月 3 日,创始合伙人张倩在厦门发表 30 页主题演讲《开源破局》,前 5 页公开,相关报告同步补上注明日期的「天际立场」小节(已证实)。