The three-year valuation rose from 300 million to 35 billion. How did Kimi do that?
Author: Beating
Original title: Kimi completed financing of more than 3.5 billion US dollars, and the valuation rose to 35 billion US dollars. The pre-IPO round has already started early
Kimi has just released K3's model weight, and a new round of funding for Dark Side of the Moon has also been settled.
According to an exclusive report from the “Science and Technology Innovation Board Daily” reporter, Dark Side of the Moon has completed Series F financing. The financing amount exceeds 3.5 billion US dollars, and the post-investment valuation has reached 35 billion US dollars.
There were no plans to raise this much money in this round of financing. The report said that Dark Side of the Moon closed the F round early because the investors' subscription amount exceeded the original target by more than three times. Series G financing, which was originally scheduled to start in August of this year, also began early.
Round G will be the pre-IPO round before Dark Side of the Moon goes public, and the pre-investment valuation has already risen to $50 billion.
A week ago, the version that came out on the market is that Dark Side of the Moon will launch the final round of private equity financing in August after completing a pre-investment round of financing valued at approximately US$31.5 billion. Today, the final round of round F funding is over $3.5 billion, and the post-investment valuation has reached $35 billion, and the next round won't wait until August. Both the money and the valuation came faster than originally planned.
Since its establishment in 2023, Dark Side of the Moon has only taken more than three years to push its valuation from $300 million to $35 billion. The ongoing pre-IPO round has put the next price tag at 50 billion US dollars.
More than ten rounds of financing in three years, and the valuation rose from US$300 million to US$35 billion
The Dark Side of the Moon was founded in April 2023 by Yang Zhilin, Zhang Yutao, Zhou Xinyu, and Wu Yuxin.
The company completed an angel round of financing of more than $200 million about two months after its establishment, and the post-investment valuation was approximately $300 million. For a big model startup that hasn't officially launched a product, this is already a rare early-stage financing.
What really brought the dark side of the moon to the center of the capital table was the A+ round of financing completed in February 2024.
The scale of this round of financing exceeded 1 billion US dollars. Alibaba led the investment, with the participation of institutions and industrial capital such as Sequoia China, Xiaohongshu, Meituan, etc., and the valuation of Dark Side of the Moon was pushed to about 2.5 billion US dollars. Half a year later, the Series B round surpassed $300 million, and the company's valuation continued to rise to $3.3 billion.
However, after the 2024 round of financing, the Dark Side of the Moon's financing pace slowed down for a while.
Over the year, China's big model market has changed markedly. Major companies such as Byte and Ali continue to lower model prices. DeepSeek is rapidly rising with open source and cost efficiency, and competition for general chat products is also gradually shifting from user growth to model capabilities, reasoning costs, and commercial revenue. Kimi used to take the lead with long texts, but with just one product label, it's already difficult to support higher valuations.
By the end of 2025, Dark Side of the Moon completed Series C financing of 500 million US dollars, and the post-investment valuation reached 4.3 billion US dollars. Since then, the company's financing has accelerated markedly.
In the first two months of 2026, Dark Side of the Moon completed multiple rounds of financing, and the valuation rose from the previous 4.3 billion US dollars to 10 billion US dollars, and further reached 18 billion US dollars.
In May of this year, the company also completed Series D financing of about 2 billion US dollars, and the post-investment valuation reached 20 billion US dollars. Participants are no longer just internet companies and market-based investment institutions; China Mobile, Guozhi Investment, CPE Yuanfeng, and several state-owned background funds have also begun to enter the shareholder list.
Round D has just been completed, and a new round of funding will commence in June. At that time, the pre-investment valuation reported in the market had reached 31.5 billion US dollars. Today, the round has finally ended with over $3.5 billion in funding, and the post-investment valuation of Dark Side of the Moon has risen to $35 billion.
Looking back at this financing curve, the most obvious change occurred in the past six months.
At the end of 2025, Dark Side of the Moon was still valued at $4.3 billion. More than half a year later, that figure has reached $35 billion, a seven-fold increase. If the next round is completed according to the $50 billion pre-investment valuation, the valuation of Dark Side of the Moon will increase by more than 160 times in more than three years.
The reasons given by capital are also becoming more straightforward. In the past, investors were betting on Yang Zhilin and a Tsinghua technical team; now they are betting on whether Dark Side of the Moon can become one of the few companies in China to stay on the cutting edge of the global model table.
After K3, the dark side of the Moon returns to the center of the table
The timing of the early closure of this round of financing came just after Kimi K3 was released.
On July 16, Dark Side of the Moon released Kimi K3. On July 27, the company further unveiled K3's full model weights, technical reports, and some critical infrastructure technology.
K3 is a hybrid expert model with a total number of 2.8 trillion parameters and 104 billion parameters per inference. It supports 1 million token contexts, and also has the ability to visualize, program, reason, and call tools over time.
The Dark Side of the Moon stated in the technical report that K3 has reached cutting-edge levels in tasks such as long-range programming, agents, knowledge, reasoning, and vision. Its overall capabilities are already very close to the strongest closed-source models, Claude Fable 5 and GPT-5.6 Sol.
The list changes all the time, and model companies' own tests also need to be distracted. The more important impact of K3 is that it pushes the power boundaries of the open weighting model a bit further.
For a long time in the past, the open source model took on more of the role of a catcher. Closed Source is responsible for training the strongest models. After half a year or a year, the open source community reproduces some of these abilities at a lower cost.
K3 changed this time difference. A Chinese company directly opened up a model close to the world's most cutting-edge capabilities. Other developers can download, modify, deploy, and continue to train their own models based on it.
The model certainly didn't get cheaper as a result. The K3's weight file is over 1.5 TB, and even just loading the full model requires multiple high-end acceleration cards. For production deployment, hardware investment may be high. Its “openness” mainly addresses whether enterprises can master the model and whether they can get rid of a single API vendor; it doesn't mean that everyone can run it on their home computer.
Even so, open weight will still affect the closed source company's business.
When a powerful enough model can be downloaded, companies have the opportunity to deploy it on their own servers, keep the data in-house, customize the model, and not have to pay for a certain company's token price forever. The barriers established by the closed source model in the past based on ability leadership will also be diluted more quickly.
Therefore, after K3 was released, the controversy quickly surpassed the technology list and entered into a dispute over the US AI industry's policy and commercial interests.
Behind the open source and closed source dispute
On July 24, companies and organizations including OpenAI, Google, Microsoft, Nvidia, AMD, Meta, and Hugging Face publicly supported the open weighting model and opposed the US government's rush to limit AI models that can be downloaded and deployed.
Anthropic and Amazon did not appear on the signing list. Outsiders immediately questioned whether Anthropic wanted to use the name of safety and national security to restrict the development of an open model and block competitors for its closed source business.
On the day of K3's opening weights, Anthropic CEO Dario Amodei published an article responding to the controversy.
He said that Anthropic has never advocated a complete ban on open weighting models. An open model without dangerous capabilities is a public product that can reduce usage costs and create value for enterprises, developers, and researchers.
However, Amodei's core position has not changed.
He still advocates continuing to restrict the flow of advanced chips and chip manufacturing equipment to China, crack down on so-called “industrial-scale model distillation,” and requires models that are sufficiently capable to undergo mandatory safety tests before release, regardless of whether these models are open or closed source.
This debate ostensibly revolves around safety; underneath it, there is also a very real business issue.
Closed Source wants to continue to control the strongest model and charge fees through APIs and subscription services; the open model allows capabilities to spread to more companies, lowers prices, and shortens the distance between latecomers and leading companies. The faster it is opened, the harder it is for a few companies to keep their model capabilities locked in their own servers for a long time.
K3 comes from China, is large enough, and its capabilities are close to the cutting edge. For those who support an open model, K3 is proof that an open route can still produce a best-in-class model; for Anthropic, it also means that the capability advantages that US companies have spent huge sums of money to build may spread at a faster rate.
As a result, the dark side of the Moon received a kind of capital value that it did not have before.
It's no longer just an AI app company with a large number of Chinese users, nor is it just a basic model startup waiting for the business model to mature. After K3 was released, it began to be re-valued in multiple competition between global open source and closed source, the US and China, model power and computing power control.
The capital market answers the $3.5 billion financing and $35 billion valuation.
Twitter:https://twitter.com/BitpushNewsCN
Compare the TG exchange group:https://t.me/BitPushCommunity
Compare TG subscriptions:https://t.me/bitpush
