On the afternoon of July 29, a flash from the Science and Technology Innovation Board Daily detonated AI venture capital circles: Moonshot AI, the company behind Kimi, had closed over $350 million in Series F funding at a $35 billion post-money valuation.

The more explosive details came later: market subscription demand for this round far exceeded expectations, reaching over 3x the original target, so the round closed early. And the Series G (Pre-IPO round) originally planned for August, according to multiple sources, is already underway.

What does $3.5 billion mean? It's one of the largest single private financing rounds in Chinese AI history. What does a $35 billion valuation mean? On the global coordinate system, it has entered the first tier of LLM companies — OpenAI is valued at roughly $200 billion, Anthropic at around $380 billion, xAI at about $80 billion. Kimi went from zero to $35 billion in less than three years.

But what deserves more attention than the numbers is the signal itself: the Pre-IPO window for Chinese LLM companies has officially opened.

Three Details That Reveal the Extraordinary Nature

A funding round with three unusual details worth savoring.

First detail: 3x oversubscribed, closed early. In the funding environment of late 2026, this is extremely rare. From late 2025 through the first half of 2026, primary market sentiment toward AI companies shifted from the 2023-2024 mania to rationality — investors wanted to see ARR, retention, unit economics, real willingness to pay. Pitch decks and stories no longer raised money. For Kimi to be oversubscribed 3x in this environment means investors already have considerable confidence in its commercialization metrics.

Second detail: Series G launching immediately after F, and it's Pre-IPO. Normal financing cadence is A→B→C→D→E→F→IPO. After Series F, companies typically spend 12-18 months preparing for a public listing. Kimi originally planned to start Series G in August; now that F has barely closed, word is G is already advancing — suggesting two possibilities: either business development is far outpacing projections and needs more ammunition, or the IPO timeline is closer than outsiders think. Given Yang Zhilin's repeated public statements about "considering listing when conditions are ripe," the latter seems more likely.

Third detail: valuation anchored to global coordinates. $35 billion wasn't pulled from thin air. When Anthropic closed its Google-led round in May this year, it was valued at roughly $380 billion; OpenAI, backed by SoftBank and others, sits around $200 billion. Kimi's $35 billion is roughly 1/11th of Anthropic and 1/6th of OpenAI — accounting for Chinese market paying power and ARPU differences, this valuation level means international investors already treat Kimi as the most important LLM play in the Chinese market, giving it a clear position in global AI asset allocation.

One easily overlooked backdrop: this round comes merely three weeks after Kimi K3's open-source release shocked Silicon Valley. When K3 launched in early July — a 2.8-trillion-parameter model, the largest open-weight model globally — it triggered massive reactions in the US: 200+ companies co-signed a letter opposing potential US government bans on open-source models, with giants like Nvidia, Microsoft, and Meta publicly supporting open AI. K3's release essentially handed global investors the hardest possible technical endorsement.

From 'Hundred Models War' to 'Tier Differentiation': Chinese LLMs Enter Elimination Rounds

Kimi's round, viewed against the broader industry backdrop, marks Chinese LLMs' shift from the "hundred models war" to "tier differentiation."

2023-2024 was the heated battle phase for Chinese LLMs: Baidu, Alibaba, Tencent, ByteDance, Huawei, Zhipu AI, Moonshot AI, MiniMax, 01.AI, DeepSeek… dozens of companies released LLMs in succession, and the "hundred models war" lived up to its name. Competition logic then was "who launches first, who has more parameters, who has higher benchmarks," and funding logic was "bet on teams, bet on tech, bet on stories."

But entering 2026, the rules changed completely. Investors no longer pay for parameters and benchmarks; they ask: how many real paying users do you have? What's your ARR? What's retention like? Have you figured out unit economics? What's your penetration in vertical scenarios?

Under the new standards, tier differentiation is already clear: First tier — Moonshot AI (Kimi), Zhipu AI (GLM series), DeepSeek, ByteDance (Doubao) — these either have ARR breaking $100M, massive owned traffic, or irreplaceable technical differentiation; Second tier — Baidu Wenxin, Alibaba Tongyi, Tencent Hunyuan — backed by giant ecosystems but independent commercialization paths still being validated; Third tier and below — facing increasingly severe survival pressure: get acquired, go vertical, or fade away.

Kimi's $350M Series F is essentially primary market voters using real money to say: in this elimination tournament, Kimi has earned its ticket to the final round.

After Pre-IPO: The Real Test Has Just Begun

Funding is a milestone, but by no means the finish line. For Kimi and other Chinese LLM companies sprinting toward IPO, the road after Pre-IPO is harder than going from zero to $35 billion.

The first test is commercial sustainability. Private market valuations can be propped up by stories and expectations, but public market valuations require hard financial numbers. The gold standard for SaaS IPOs is "120% NDR (Net Dollar Retention) + 20%+ free-to-paid conversion + a clear path to profitability." What exactly Kimi's ARR, retention, and gross margins are, outsiders don't know — but to support a $35 billion or higher IPO valuation, those numbers must withstand scrutiny from Wall Street (or HKEX / the STAR Market).

The second test is technical moats. LLM technology iterates on a monthly basis — today you release the strongest model; three months later competitors may catch up or surpass you. K3's 2.8 trillion parameters is impressive, but open-source models' nature means everyone can learn from, reference, and even improve upon it. How to maintain leadership in a fast-moving technology race is the ultimate question for every LLM company. OpenAI built its moat on closed-source + compute + ecosystem; Anthropic built its moat on safety + enterprise customers. What is Kimi's moat? Long context? Open-source community? Consumer product experience? This question needs a clear answer before IPO.

The third test is geopolitical risk. This is a problem no Chinese AI company can bypass. The US has already imposed export controls on high-end AI chips to China; after Kimi K3 went open-source, over 200 American companies co-signed opposition to a ban — indicating divisions within US policy circles on how to treat Chinese open-source models — but risk always lingers. Chip supply, overseas market access, cross-border data flows — each is a variable that could impact valuation.

The fourth, most easily overlooked test: organizational capability upgrade from startup to public company. In the private phase, companies can tolerate a degree of chaos and trial-and-error, but after listing, financial compliance, corporate governance, information disclosure, investor relations — every link must be error-free. OpenAI's biggest pre-IPO challenge wasn't technology but corporate governance structure (complexity of nonprofit→for-profit transition). Chinese AI companies will face the same question.

Verdict: Chinese AI's 'Capital Moment'

Pulling back from Kimi as a single company, you'll find a larger trend forming: late 2026 through first half of 2027 will be a concentrated IPO window for Chinese AI companies.

Zhipu AI is reportedly preparing a Hong Kong listing, MiniMax is rumored to be talking to investment banks, and if ByteDance spins off its AI business for an IPO, the valuation could be astronomical — LLM companies have raised enough rounds and burned enough money in private markets; investors need exit channels, and companies need the sustained funding capacity of public markets to support the compute arms race.

This follows the same logic as the concentrated listings of Alibaba and JD.com around 2014, and Xiaomi, Meituan, Pinduoduo around 2018: when an emerging industry moves from the VC stage to secondary markets, it marks its transition from "emerging" to "mature," from "story" to "performance."

Of course, risks are real. The lessons of the 2021-2022 US SaaS bubble burst are still fresh — a cohort of high-growth SaaS companies saw share prices drop 70-90% post-IPO when public markets discovered their "high growth" came at the cost of "massive losses." If AI companies repeat that pattern, it won't just be primary market investors who get hurt — it will be confidence in the entire industry.

But regardless, Kimi's $350M Series F is a moment worth marking. It means the Chinese LLM industry has truly walked under the global spotlight; the IPO race ahead will be an all-around competition of technology, commercialization, organizational capability, and geopolitics.

$3.5 billion isn't the finish line — it's the starting gun. When Chinese LLM companies hit $35 billion valuations, the real race has just begun.

See you tomorrow.