China's Moonshot AI is targeting up to $12B valuation in a $700M funding round—179% jump from $2.5B...
The AMW Read
The article validates Frame 2 of the Foundation Model debate by demonstrating open-weight models setting the pace against proprietary labs, while meeting cross.§D and cross.§B triggers via the $700M round and explicit open-weight scaling strategy.
NoveltySignificance
Foundation Models · Player MapScaling LawsCapital Cycles
China's Moonshot AI is targeting up to $12B valuation in a $700M funding round—179% jump from $2.5B just one year ago—driven by Kimi K2.5's benchmark wins over GPT-5.2 and Claude Opus 4.5. The 1T-parameter open-weight model outperformed proprietary competitors on Humanity's Last Exam (50.2% vs 45.5%), while overseas revenue surpassed domestic income for the first time with 4x international user growth. This signals a structural shift: open(-weight) models are no longer catching up to closed labs—they're setting the pace. With Zhipu AI and MiniMax already valued at ~$29B post-IPO, investor appetite for China's AI race remains voracious despite US chip restrictions.
