The Dark Side of the Moon and other major AI companies compete for IPO: From technological rivalry to a new battlefield of capital valuation

The Chinese large model industry is undergoing a critical shift, with leading companies shifting their strategic focus from primary market financing to secondary market listings. The direct trigger for this change was the successful listings of Zhipu and MiniMax on the Hong Kong Stock Exchange earlier this year, with both companies’ market values surpassing HKD 300 billion in a short period, and their stock prices rising 3 to 5 times from their issuance prices, providing the first public valuation benchmark for the industry. The valuation system, originally reliant on private equity financing plays, is beginning to be replaced by the pricing logic of the public market.

The movements of Moonlight Dark Side are the most representative. Three months ago, the company emphasized its strategic determination of “not going public in the short term,” but now it has initiated preparations for an IPO in Hong Kong and has begun discussions with investment banks such as CICC and Goldman Sachs. According to industry insiders, its C+++ round financing scale reached USD 1 billion, with a pre-investment valuation jumping to USD 17-18 billion, achieving a threefold increase compared to the USD 4.3 billion at the end of last year. This valuation leap is not merely a result of corporate growth, but also reflects the transmission effect of secondary market pricing on the primary market—according to the benchmark logic of Anthropic, its long-term market value target even points to USD 38 billion.

The new rules of the capital market are reshaping the industry’s competitive landscape. StepStar is simultaneously advancing a new round of financing with a valuation of USD 6 billion, planning to complete the deal by mid-April, with cornerstone pricing expectations reaching USD 10 billion. This “battle for the third listed company” is essentially a scramble for entrance to the capital market. Whoever completes the listing first will gain pricing power, becoming the valuation reference for subsequent companies. This locking effect is common in the tech industry, where early public companies often consolidate their technological lead through capital advantages.

The core driving force behind this capital race is the substantial breakthrough in the large model business model. Taking Moonlight Dark Side as an example, its K2.5 model achieved revenues exceeding the total for the year 2025 within 20 days by enhancing multi-agent scheduling capabilities. This shift stems from a restructured demand framework: moving from low-frequency dialogue scenarios to high-frequency task execution scenarios, where a single complex task can be broken down into hundreds of sub-tasks, leading to a surge in calling frequency and token consumption. On the OpenRouter platform, the token consumption share of Chinese models has exceeded 60%, with Kimi K2.5 ranking seventh globally with a monthly consumption of 30 trillion tokens.

The evolution of the revenue model has directly altered market assessment standards. Moonlight Dark Side has actively contracted its C-end pan-entertainment business, concentrating resources on the development of foundational models, resulting in a drop in MAU from 36 million to 9.67 million, yet valuation and revenue have risen against the trend. This “de-user scaling” strategy reflects a market re-recognition of “unit calling value.” More crucially, its overseas revenue share has for the first time surpassed that of the domestic market, with the number of personal subscription user payment orders skyrocketing nearly a hundred times within two months, and global payment rankings soaring to ninth place, indicating that Chinese large models are integrating into the global developer ecosystem.

The pricing logic of the capital market has also been upgraded. Li Rui, Executive Director of Qishijie Technology, pointed out that large model companies are transitioning from technology organizations to capital assets, with assessment standards shifting from model parameters to sustainable revenue capabilities, growth paths, and liquidity levels. For companies like Moonlight Dark Side, which has cash reserves exceeding RMB 20 billion, the core purpose of going public is no longer financing, but rather converting book valuations into public market prices, providing a cash-out channel for equity incentives and constructing a positive cycle of continuous financing.

This transformation is triggering a chain reaction. When a company completes its listing first and establishes stable pricing, subsequent companies will be compared across various dimensions such as financing, talent, and market perception within the same coordinate system, raising industry entry barriers. The accelerated sprint of Moonlight Dark Side and StepStar is essentially a proactive adaptation to the new competition rules—in a technology race that requires continuous high investment, those who can enter the capital cycle system earlier will be able to convert short-term advantages into long-term barriers.

Currently, the IPO fundraising scale of the Hong Kong stock technology sector has surged by 300% year-on-year, with AI becoming the core driving force. This dual variation of capital and technology marks the entry of Chinese large model competition into a new phase: technology determines the lower limit of survival, while capital operation capability begins to dominate the upper limit of development. When pricing power shifts from laboratories to exchanges, the rules of the game are being completely rewritten.