Addy Crezee
moonshot ai closed $3.5b at a $35b valuation on july 29, 13 days after shipping kimi k3. it targeted $1–2b and got nearly triple that. the round is 7x the size of its december raise – at a valuation 8x higher, seven months later
@Kimi_Moonshot's ladder, compressed:
• jun 2023 – $200m seed at $300m
• dec 2025 – $500m series c at $4.3b
• jul 2026 – $3.5b at $35b
that's 117x on valuation and 17x on round size in 37 months. $7.9b of the $9.4b it has ever raised arrived in the last 8 months
observations:
1. the model didn't start the round, it resized it – the raise opened in june at a $31.5b pre-money, targeting $1–2b. k3 shipped july 16, and it closed on july 29 at $3.5b. a launch is no longer what triggers a round, it's what closes one
2. the 16-month freeze is the tell – moonshot shipped k1.5 and k2 between aug 2024 and dec 2025 and raised nothing, frozen by an arbitration case brought by investors in yang zhilin's previous company. strong models don't raise money on their own
3. it's the revenue, not the benchmark – arr went $100m in march, $200m in april, $300m in june. after k3 daily sales rose at least 6x and moonshot had to stop taking new subscriptions on compute limits
4. the state showed up – the national ai industry investment fund, also a deepseek backer, led. china's frontier labs are shifting from alibaba/tencent money to state capital as the checks get bigger
5. it isn't done – moonshot is already talking to investors at a $50b pre-money for a final private round, with a hong kong listing targeted as soon as this year. another 43% on top, before it prices publicly
