TL;DR: Moonshot AI closed a $3.5 billion funding round at a $35 billion valuation on July 29, blowing past its initial $1-2B target. The Beijing-based lab is already in talks for another round at a $50 billion pre-money valuation and plans a Hong Kong IPO as soon as this year — all riding the momentum of Kimi K3, the 2.8 trillion-parameter open-weight model that beat Fable 5 and GPT-5.6 Sol on multiple coding benchmarks two weeks ago.
The Oversubscribed Round That Wasn’t Supposed to Happen
When Moonshot AI first started fundraising in June, the target was modest by 2026 AI standards: $1 billion to $2 billion at a roughly $30 billion valuation (Source: Bloomberg — China’s Moonshot AI Seeks $30 Billion Value in New Funding Talks). What actually closed on July 29 was a $3.5 billion round at $35 billion — nearly double the upper bound of the original ask (Source: Bloomberg — Moonshot AI Surpasses Funding Goal to Hit $35 Billion Value).
The round was driven entirely by the Kimi K3 launch on July 16. Investors who had been cautious about Chinese AI labs suddenly wanted all the allocation they could get. Moonshot didn’t just meet its target — it had to turn money away.
And it’s not stopping. Bloomberg reports the company is already in talks for a $50 billion pre-money round, aiming to secure one final private capital injection before filing for an IPO in Hong Kong later this year (Source: Bloomberg — Moonshot in Talks on Pre-IPO Funds at $50 Billion Value). That would put Moonshot’s valuation curve at roughly a 14x increase in 18 months: $2.5B (Feb 2024) → $3.3B (Aug 2024) → $35B (Jul 2026) → $50B+ (pre-IPO, expected late 2026).
The Kimi K3 Catalyst
The financial story is inseparable from the technical one. Kimi K3 landed on July 16 as the largest open-weight AI system ever released — 2.8 trillion parameters in a Mixture-of-Experts architecture, with a 1 million token context window and only 16 of 896 experts activated per token.
The benchmarks tell the story investors paid attention to:
- #3 on the Artificial Analysis Intelligence Index, trailing only Claude Fable 5 and GPT-5.6 Sol by a few points
- #1 on Frontend Code Arena, beating both Fable 5 and GPT-5.6 Sol
- #1 on Next.js code generation and migration tasks
- #1 on Arena WebDev leaderboard at launch
- Terminal-Bench 2.1 at 88.3, DeepSWE at 67.5, FrontierSWE at 81.2
(Source: OfficeChai — Kimi K3 Beats Fable 5 & GPT 5.6 Sol on Frontend Code Arena, Artificial Analysis Intelligence Index)
Moonshot open-sourced the weights on Hugging Face alongside a technical report, creating what the platform described as its fastest-growing model launch ever — thousands of likes within 30 minutes. For investors, the combination of frontier-level performance PLUS open-weight availability PLUS roughly half the API cost of closed competitors added up to a narrative too compelling to ignore.
We covered the K3 launch in detail on July 17: /2026/07/kimi-k3-moonshot-28t-open-model-july-2026/.
The Geopolitical Shadow
The funding euphoria exists alongside serious geopolitical friction. The White House has accused Moonshot of training Kimi K3 on restricted Nvidia chips obtained in violation of U.S. export controls, and has alleged the model was built partly through “distillation” — extracting outputs from rival AI systems (Source: Stocktwits — Moonshot AI 35B Valuation).
The Information separately reported that Moonshot is seeking access to Nvidia’s Blackwell processors to train Kimi K4, the successor to K3. This puts the company in a delicate position: raising billions from global investors while navigating an increasingly restrictive U.S. export control regime.
We explored the White House distillation accusations on July 24: /2026/07/moonshot-kimi-k3-white-house-distillation-accusation-july-2026/.
The Nvidia angle matters to the broader market story too. Kimi K3’s release contributed to a chip stock selloff that wiped hundreds of billions from Nvidia’s market cap, knocking it from its position as the world’s most valuable company — a title it lost to Apple amid the rout. Nvidia fell from a peak of $5.1 trillion to roughly $4.7 trillion.
The Chinese AI Gold Rush
Moonshot isn’t operating in a vacuum. Chinese AI labs are racing toward Hong Kong listings at valuations that would have seemed absurd 12 months ago. Z.AI’s stock is trading at nearly 10x its IPO price from January 2026. MiniMax has filed confidentially for its own Hong Kong listing. Chinese open-weight models now account for roughly 30% of global usage, up from just 1.2% eleven months ago (Source: ThursdAI — July 2026 AI Releases).
The market is treating these companies less like DeepSeek — a research lab that happened to release great models — and more like the early days of the US cloud giants. The thesis: if open-weight Chinese models can match or beat closed frontier models at half the cost, the addressable market for API inference and enterprise deployment is enormous.
Moonshot’s IPO plans were detailed on July 20: /2026/07/moonshot-ai-hong-kong-ipo-20-billion-kimi-k3/.
FAQ
Q: Who invested in the $3.5B round? Moonshot hasn’t disclosed the full investor list. Previous rounds included Alibaba, Sequoia China, Tencent, and Gaorong. The new round’s participants haven’t been named yet.
Q: How does Moonshot’s valuation compare to other AI labs? At $35B, Moonshot is approaching Anthropic’s most recent private valuation range (estimated $40-60B) and has surpassed Mistral (~$6B) and Cohere (~$5.5B). It trails OpenAI (~$300B) by a wide margin but is closing the gap with the second tier of US labs faster than anyone predicted.
Q: Will export controls derail the IPO? That’s the trillion-dollar question. If the U.S. blocks Moonshot’s access to Blackwell chips, Kimi K4 could be significantly constrained. But Moonshot has demonstrated it can train frontier models on existing hardware — and the IPO filing will force unprecedented transparency about its chip sourcing and model training practices.
Q: When is the Hong Kong IPO expected? Bloomberg reports “as soon as this year,” meaning late 2026. Moonshot is raising the $50B pre-IPO round first, which suggests the filing timeline is measured in months, not quarters.
Q: Is the $35B valuation justified? By revenue multiples, almost certainly not — Moonshot’s commercial revenue is still modest compared to its valuation. But AI lab valuations in 2026 are driven by model capability, not current revenue. Kimi K3’s benchmark performance places it in the same tier as models from companies valued at 3-8x more, which makes $35B look like a relative bargain if you buy the thesis that open-weight frontier models will capture significant API market share.
Further Reading
- Bloomberg — Moonshot AI Surpasses Funding Goal to Hit $35 Billion Value
- Bloomberg — Moonshot Plans IPO in Six Months After AI Breakthrough
- OfficeChai — Kimi Maker Moonshot AI Valued at $35 Billion After $3.5 Billion Fundraise
- Stocktwits — The Chinese AI Model That Wiped Hundreds of Billions Off Nvidia
- ThursdAI — July 2026 AI Releases (54 launches)
- TAR — Kimi K3: Moonshot AI Drops a 2.8T Open Model
- TAR — Moonshot AI Plans Hong Kong IPO at $20B+ Valuation
- TAR — White House Accuses Moonshot of Training Kimi K3 on Restricted Chips