Revenue Doubles, What Signal Does MiniMax's First Financial Report Release?

Who Can Outperform AI in the Second Half?

On March 10, 2026, a milestone was reached in the large AI model industry. MiniMax (0100.HK) stock price closed up over 22%, with a total market capitalization rising to HKD 382.64 billion, officially surpassing Baidu (market cap HKD 332.22 billion). This reversal in market value signifies strong recognition from the capital market of the commercial potential of pure AI large models.

Chart: Closing prices of MiniMax and Baidu in Hong Kong stocks on March 10

Impressive Performance: Revenue Doubled, Loss Structure Analyzed

As the first annual report since its listing on January 9, 2026, MiniMax’s audited financial results for 2025 show a strong growth trend. In 2025, the company’s revenue increased by 158.9% year-over-year to $79.038 million. The financial report indicates a loss of $1.872 billion under certain accounting standards, but adjusted net loss was $250 million, with the loss margin remaining manageable.

In terms of profitability, gross margin rose to 25.4%, up 13.2 percentage points year-over-year. The company attributes this to improvements in model and system efficiency and infrastructure optimization. Regionally, overseas markets have become growth engines, contributing 73% of revenue; domestic Chinese revenue also maintained rapid growth. By product, revenue from AI-native products accounts for over 60%, and revenue from open platforms and enterprise services surged 197.8% year-over-year, indicating a clearer commercialization path.

Technological Moat: Multimodal Capabilities and Agent Deployment

On the technical front, MiniMax has built full multimodal R&D capabilities. Its model family includes text models M2, video models Hailuo2.3, speech models Speech2.6, and music models Music2.0. In terms of language models, its M2 model achieved over 50 billion tokens consumed daily on the OpenRouter platform in November 2025, becoming the first Chinese model to reach this milestone. According to late February 2026 data, its iterative model M2.5’s weekly token consumption on OpenRouter has ranked among the top globally.

More notably, the internal deployment of AI Agents is gaining traction. Currently, MiniMax’s internal Agents are integrated into many employees’ workflows. The company predicts that fields like programming may advance toward L4 (strategic autonomous agents) and even L5 (fully creative), enabling a leap from “tools” to “colleague-level” collaboration.

Industry Outlook: From “Burning Money for Training” to “Value Realization”

In recent years, the AI large model industry has gone through a phase of “burning money for technology,” with companies investing heavily in training models and data accumulation, but with unclear commercialization paths. However, as model efficiency improves, Agent technology is deployed, and enterprise services deepen, the industry is gradually shifting toward a new stage emphasizing “efficiency-driven” and “scenario-based” applications. MiniMax’s performance (such as gross margin improvement, overseas growth, and enterprise client expansion) exemplifies this trend.

Government policies supporting the AI industry and capital market attention to AI large model companies have created a favorable environment for development. As one of the first AI large model companies in Hong Kong stocks to surpass HKD 100 billion in market value, MiniMax’s surpassing Baidu also reflects a reassessment of the value of AI-native companies by the capital market. With more policy support and capital inflow, the industry is expected to accelerate into a “value realization” phase.

MiniMax’s market cap surpassing Baidu may just be the beginning of a revaluation of AI-native enterprise value. In the future, whoever achieves large-scale commercialization of Agents first will dominate the next round of competition.