Following one another, AI enterprises are debuting on the Hong Kong stock market—first came Knowledge Atlas Tech, and now MINIMAX-WP (00100.HK).

Since the beginning of 2026, despite the downward volatility of the Hang Seng Tech Index, these two AI "bull stocks" have surged against the trend. They once broke through the HK$300 billion market capitalization milestone, surpassing mature internet platform companies such as Trip.com and JD.com.

Although MINIMAX's share price has recently fluctuated and retreated from its highs, the company’s market capitalization remains firmly above HK$200 billion.

P/S Ratio Exceeds 400x, Significantly Higher Than Global AI Giants

Public records show that MINIMAX was founded in 2022. From its inception, it has focused on the research and development of "all-modal" models. As a globalized AI large-model company, it is committed to driving technological innovation in artificial intelligence to achieve the capability to execute various human intellectual tasks, including learning, reasoning, planning, and generalizing knowledge across diverse fields.

MINIMAX disclosed in its prospectus that it adopted a Mixture of Experts (MoE) architecture and a hybrid attention mechanism in its early stages. This allows the company to maintain world-leading performance while significantly reducing computational resource consumption.

NewTimeSpace has learned that in January 2024, MiniMax released China’s first MoE large model. In April of the same year, it began deploying the MoE architecture for large-scale commercial use.

According to the prospectus, the entire reinforcement learning stage of the MiniMax M1 model required only 512 H800 GPUs for three weeks of training. The training cost was US$537,400, which is merely a fraction of the cost for an OpenAI model with equivalent parameters. By the time of its listing, MiniMax’s cumulative spending was approximately US$500 million, only 1% of OpenAI’s spending over the same period.

MINIMAX was greeted with intense enthusiasm from capital upon its listing. In February 2026, the company’s share price once surged to HK$980. Given an offer price of HK$165, the maximum gain since listing reached nearly sixfold. As of the close on February 27, MINIMAX was quoted at **HK$763.5**. Although it has retreated from its peak, it still represents a nearly fourfold increase from its offer price, with a market capitalization of HK$239.5 billion.

NewTimeSpace has observed that following this massive surge in share price, the company’s Price-to-Sales (P/S) ratio has become significantly higher than that of current global AI giants.

While MINIMAX has not yet released its full-year results for 2025, the latest forecast data from CICC estimates MINIMAX’s revenue to be US$71 million, compared to US$53 million for the first three quarters of 2025.

Based on MINIMAX’s closing market cap on February 27 (approximately US$30.616 billion), its P/S ratio stands at a staggering 431.21 times.

For comparison, NewTimeSpace understands that OpenAI, currently valued at US$830 billion with a forecasted 2025 revenue of US$13 billion, has a P/S ratio of approximately 63.85 times. Anthropic, valued at US$380 billion with a forecasted 2025 revenue of US$4.7 billion, corresponds to a P/S ratio of 80.85 times.

Knowledge Atlas Tech, which listed one day before MINIMAX, has a P/S ratio of 314.56 times, based on CICC’s estimated 2025 revenue of US$104.2 million and its February 27 closing market cap of HK$256.4 billion (approx. US$32.777 billion).

Revenue analysis is a crucial step in assessing a company’s business prospects. Generally, a lower P/S ratio indicates greater investment value. Currently, the investment value of both Knowledge Atlas Tech and MINIMAX appears significantly discounted given their current revenues relative to their market capitalizations.

C-End + B-End Layout: Building an All-Modal Product Ecosystem

NewTimeSpace has learned that MINIMAX primarily focuses on C-end (consumer) products based on multi-modal technology, targeting both domestic and international markets simultaneously.

The company’s current revenue streams consist of two parts: AI-native products for C-end users, and an open platform alongside other AI-based enterprise services.

In the C-end market, MiniMax has deployed two major products. The AI social application Talkie/Xingye was launched early on. Supported by MiniMax’s multi-modal technology, users can engage in chats and calls with AI characters.

It is worth noting that MINIMAX’s products were globalized from the start. The company launched Talkie for the international market first; after accumulating a user base overseas, it introduced the domestic version, Xingye.

In 2024, based on its video and music models, MiniMax launched the AI video generation app Hailuo AI. In June 2025, the video generation model Hailuo 02 was released, ranking second globally in video evaluations by the platform Artificial Analysis, trailing only ByteDance’s Seedance 1.0. In October, Hailuo 2.3 was released.

With its technical foundation, Hailuo AI has become the second growth curve for MiniMax in the C-end market.

On the B-end (business), MiniMax provides API call services to enterprise customers and developers through the MiniMax Open Platform. To date, MiniMax has constructed an all-modal product system covering text, voice, and video.

Currently, MINIMAX supports multi-modal content creation through Hailuo AI (Hailuo AI), while Xingye/Talkie focuses on AI emotional companionship. Its voice model has generated a cumulative total of over 220 million hours of content, and its video model has generated over 590 million videos.

Furthermore, around the Lunar New Year holiday, MINIMAX released its next-generation model, M2.5. Within 12 hours of its release, it topped the OpenRouter popularity chart. Within a week, it led the charts in invocation volume, with weekly volume skyrocketing to 3.07T tokens—exceeding the combined total of Kimi K2.5, GLM-5, and DeepSeek V3.2.

High R&D Investment: Cumulative Losses Exceed RMB 9 Billion Since Inception

The prospectus shows that MiniMax’s revenue was US$3.46 million in 2023, increasing to US$30.52 million in 2024. Revenue for the first nine months of 2025 reached US$53.44 million, a 175% increase year-on-year. 70% of the company’s revenue comes from overseas markets. Currently, the company has 212 million individual users and 130,000 enterprise customers, with paid users exceeding 1.77 million, demonstrating strong consumer internet characteristics.

Recently, Morgan Stanley initiated coverage on MiniMax with an "Overweight" rating and a price target of HK$930, positioning it as a "Global AI Foundation Model Leader." However, Morgan Stanley expects that while the gross margin will improve from 12% in 2024 to 32% by 2027, operating losses will widen during the same period. The non-IFRS operating loss for 2027 is estimated at approximately US$484 million.

Listing is not the finish line, but the start of a new race. Under the intense competition of domestic and international large models, MINIMAX faces rivalry not only from Knowledge Atlas Tech—which is also listed in Hong Kong—but also from overseas giants and AI models launched by major internet firms.

In its IPO prospectus, MINIMAX disclosed an estimated monthly cash burn of US$27.9 million. From 2022 to 2024, MiniMax recorded net losses of US$73.73 million, US$269 million, and US$465 million, respectively. The net loss for the first three quarters of 2025 reached **US$512 million**, already surpassing the total for the full year of 2024. From its founding to its listing, the company’s cumulative loss has reached US$1.32 billion (approx. RMB 9.053 billion).

Specifically, MINIMAX’s cumulative R&D expenses since 2022 total US$450 million (RMB 3.086 billion). Of this, approximately US$330 million was spent on training-related cloud computing services, accounting for 74% of total R&D costs.

Currently, the large-model sector is a classic high-investment industry. Product competition is essentially a competition between underlying models. The capabilities of these underlying models must be maintained through continuous R&D training and iterations. With the rapid pace of AI technology evolution, model manufacturers must sustain high-intensity R&D investment just to stay in the game.

Much like Knowledge Atlas Tech, which also faces massive losses, MINIMAX needs to find a balance between profitability and R&D investment. It also needs to verify whether its product-driven model can support long-term technological input, turning large models into a sustainable business that justifies a market capitalization exceeding HK$200 billion.

NewTimeSpace Disclaimer: All content herein is the original work of NewTimeSpace. Any reproduction, reprinting, or use of this content in any other manner must clearly indicate the source as "NewTimeSpace". NewTimeSpace and its authorized third-party information providers strive to ensure the accuracy and reliability of the data, but do not guarantee the absolute correctness thereof. This content is for reference only and does not constitute any investment advice. All transaction risks shall be borne by the user.