Kimi has just released the model weights for K3, and a new funding round for the Dark Side of the Moon has been finalized.
According to the exclusive information from the "Science and Technology Innovation Board Daily," the Dark Side of the Moon has completed its Series F funding round, raising over $3.5 billion, with a post-money valuation of $350 billion.
This funding round was not originally intended to raise this much capital. Reports indicate that due to investors subscribing for an amount more than three times the original target, the Dark Side of the Moon closed the Series F round ahead of schedule. The G round, initially planned to start in August this year, has now been brought forward.
The G round will be the Pre-IPO round for the Dark Side of the Moon, with a pre-investment valuation already raised to $500 billion.
A week ago, the market rumors suggested that after completing a pre-investment valuation round of around $31.5 billion, the Dark Side of the Moon would initiate the final private funding round in August. However, with the Series F round raising over $3.5 billion, resulting in a $350 billion post-money valuation, the next round did not wait until August. Both the funding and valuation arrived faster than originally planned.
Since its establishment in 2023, the Dark Side of the Moon has increased its valuation from $3 billion to $350 billion in just over three years. The ongoing Pre-IPO round has now set the next price tag at $500 billion.
Over a Dozen Rounds of Funding in Three Years, Valuation Soars from $3 Billion to $350 Billion
Founded in April 2023, the Dark Side of the Moon was established by Yang Zhilin, Zhang Yutao, Zhou Xinyu, and Wu Yuxin.
About two months after the company's founding, it completed a Series A funding round of over $200 million, with a post-money valuation of about $3 billion. For a large-model startup that had not yet officially launched its product, this early-stage funding was already a rare and significant amount.
What truly brought the Dark Side of the Moon to the center of the capital arena was the Series A+ funding completed in February 2024.
This round of funding exceeded $1 billion, with Alibaba leading the investment and institutions and industry capitals such as Sequoia China, Xiaohongshu, and Meituan participating. The Dark Side of the Moon's valuation was pushed to approximately $25 billion. Half a year later, the Series B funding round raised over $3 billion, further raising the company's valuation to $33 billion.
However, after the funding round in 2024, the funding pace of the Dark Side of the Moon slowed down.
During this year, significant changes occurred in the large-scale model market in China. Major players such as Byte and Alibaba continued to push down model prices, DeepSeek rapidly rose through open source and cost efficiency, and the competition for general chat products gradually shifted from user growth to model capabilities, inference costs, and business revenue. Kimi, once a pioneer in long text, now finds it challenging to support a higher valuation with just a single product label.
It wasn't until the end of 2025 that the Dark Side of the Moon completed a $500 million Series C funding round, reaching a post-investment valuation of $4.3 billion. After this point, the company's funding accelerated significantly.
In the first two months of 2026, the Dark Side of the Moon successively completed multiple funding rounds, with the valuation rising from the previous $4.3 billion to $10 billion, and further reaching $18 billion.
In May of this year, the company completed another approximately $2 billion Series D funding round, with a post-investment valuation of $20 billion. The participants were no longer limited to internet companies and market-oriented investment institutions; China Mobile, Guozhi Investment, CPE Source Peak, and several state-owned background funds also began to enter the shareholder list.
Just after the Series D round, a new round of funding was initiated in June. At that time, the rumored pre-investment valuation had already reached $31.5 billion. Today, this round ultimately concluded with a financing scale of over $3.5 billion, raising the Dark Side of the Moon's post-investment valuation to $35 billion.
Looking back at this funding curve, the most significant changes occurred in the past six months.
At the end of 2025, the Dark Side of the Moon's valuation was still $4.3 billion. A little over half a year later, this number had reached $35 billion, an increase of over seven times. If the next round is completed at the rumored $50 billion pre-investment valuation, the Dark Side of the Moon's valuation increase over a little more than three years will exceed 1,600 times.
The reasons provided by the capital are also becoming more direct. Previously, investors were betting on Yang Zhilin and a tech team with ties to Tsinghua; now, they're betting on whether the Dark Side of the Moon can become one of the few companies in China to stay at the global cutting-edge model table.
After K3, the Dark Side of the Moon returns to the center of the table
The timing of this round of funding closing happened to fall just after the release of Kimi K3.
On July 16, the Dark Side of the Moon released Kimi K3. On July 27, the company further opened up the full model weights, technical reports, and some key infrastructure technologies of K3.
K3 is a hybrid expert model with a total of 2.8 trillion parameters, each inference activating 104 billion parameters. It supports a context of 1 million tokens and has the ability for vision, programming, reasoning, and long-context tasks.
The Dark Side of the Moon stated in a technical report that K3 has achieved cutting-edge performance in tasks such as long-range programming, agents, knowledge, reasoning, and vision. Its overall capability is very close to the strongest closed-source models, Claude Fable 5 and GPT-5.6 Sol.
The rankings are subject to change, and caution is advised even with the model company's own tests. The most significant impact of K3 is pushing the capability boundary of open-weight models to a new level.
For a long time, open-source models have mostly played the role of followers. Closed-source companies were responsible for training the strongest models, and half a year or a year later, the open-source community replicated some of that capability at a lower cost.
K3 has changed this time gap. A Chinese company has directly open-sourced a model with capabilities close to the global cutting edge, allowing other developers to download, modify, deploy, and even further train their models based on it.
Of course, the model has not become cheap. K3's weight files exceed 1.5 TB, and even just loading the full model requires multiple high-end accelerator cards. For production environment deployment, the hardware investment may be significant. Its "openness" mainly addresses whether enterprises can control the model and break free from a single API supplier, rather than implying that everyone can run it on their home computers.
Nevertheless, open weights will still disrupt closed-source companies' business.
When a sufficiently powerful model can be downloaded, enterprises have the opportunity to deploy it on their own servers, keep the data in-house, customize the model, and no longer be bound to a single company's token pricing forever. The barriers that closed-source models relied on in the past to establish a lead in capabilities will also be diluted more quickly.
Therefore, after the release of K3, the debate quickly moved beyond the technical rankings and entered into a discussion of policy and commercial interests in the U.S. AI industry.
The Battle between Open Source and Closed Source
On July 24, companies and institutions such as OpenAI, Google, Microsoft, NVIDIA, AMD, Meta, and Hugging Face publicly endorsed open-weight models and opposed the U.S. government's haste to restrict AI models available for download and deployment.
Anthropic and Amazon were not part of the signatories. This led to external questioning of whether Anthropic wishes to restrict the development of open models under the guise of security and national interests to protect its closed-source business from competitors.
On the day K3 Open Weight was unveiled, Anthropic CEO Dario Amodei released an article in response to the controversy.
He stated that Anthropic had never advocated for a blanket ban on open weight models, highlighting that open models without harmful capabilities are a public good that can reduce usage costs and create value for businesses, developers, and researchers.
However, Amodei's core position remained unchanged.
He still advocated for continued restrictions on advanced chips and chip manufacturing equipment from flowing into China, opposed the concept of "industrial-scale model distillation," and called for sufficiently capable models to undergo mandatory security testing before release, whether these models are open or closed source.
While the debate appears to revolve around security, there is also a very practical business issue at its core.
Closed-source companies wish to maintain control of the most powerful models, charging fees through APIs and subscription services; open models, on the other hand, allow capabilities to diffuse to more companies, driving down prices and narrowing the gap between newcomers and leading firms. The faster the openness, the more challenging it becomes for a few companies to monopolize the long-term storage of model capabilities on their own servers.
K3, originating from China, boasts a significant scale and its capabilities are already approaching the cutting edge. For those supporting open models, K3 proves that the open approach can still yield top-tier models; for Anthropic, it also implies that the capability advantage built by U.S. companies at a significant cost may spread more rapidly.
As a result, the Dark Side of the Moon has gained a capital value it did not possess before.
It is no longer just an AI application company with a large number of Chinese users, nor is it merely a foundational model startup waiting for its business model to mature. After the release of K3, it has been revalued in a global competition involving open source and closed source, the U.S. and China, model capability, and computing power control.
The $3.5 billion funding and $35 billion valuation are the answers provided by the capital markets.
Welcome to join the official BlockBeats community:
Telegram Subscription Group: https://t.me/theblockbeats
Telegram Discussion Group: https://t.me/BlockBeats_App
Official Twitter Account: https://twitter.com/BlockBeatsAsia
