“龙虾”时代,大模型公司的好日子来了
远川研究所·2026-03-14 13:10

Core Viewpoint - MiniMax has experienced a significant stock surge, with a 51% increase over two trading days, driven by the popularity of its product OpenClaw, which has positioned it favorably against competitors like Baidu [6][7]. Group 1: MiniMax's Performance - MiniMax's stock price has risen over 600% since its IPO, with a market capitalization surpassing Baidu for the first time [6]. - The company's revenue for 2025 was approximately 540 million RMB, reflecting a year-on-year increase of 158.9% [6]. - Despite the revenue growth, MiniMax remains in a loss position as of the end of 2025 [20]. Group 2: OpenClaw's Impact - OpenClaw is a standardized framework for building intelligent agents, allowing developers to create and share various functionalities [8]. - The framework has gained immense popularity, surpassing 200,000 stars on GitHub, making it one of the fastest-growing open-source projects in history [21]. - OpenClaw's operational model significantly increases token consumption, with reports of users burning millions of tokens for simple tasks [25][27]. Group 3: Market Dynamics - The introduction of OpenClaw has created a new revenue stream for AI model companies, addressing the challenge of monetization in the AI sector [10]. - The demand for AI services is expected to grow exponentially, with predictions indicating that by 2031, Chinese enterprises will have 350 million active intelligent agents [26]. - MiniMax's annual recurring revenue (ARR) has surged from $100 million to $150 million within two months, indicating strong market confidence [29]. Group 4: Competitive Landscape - Competitors like Zhizhu and major tech companies are also launching similar products to capitalize on the OpenClaw trend, indicating a highly competitive environment [26]. - Zhizhu's AutoClaw and MiniMax's MaxClaw are examples of products designed to enhance user experience and accessibility in AI applications [26]. - The market is witnessing a shift where the focus is on attracting users to select specific models for their agents, rather than just improving benchmark rankings [28].