Is the market still pricing Jijia (02590) as a "hardware equipment company"? It is already doing "physical AI"

The 2026 World Artificial Intelligence Conference (WAIC) has just ended, and the "Human-shaped Siasun Robot & Automation Show" at various booths still left a deep impression - Yushu's human-shaped Siasun Robot & Automation G1 performed a difficult backflip, while Zhiyuan's expedition A2 demonstrated smooth human-machine interaction. But one company's booth had a completely different style. Extreme Wisdom (02590) had no dancing, no walking, no dialogue. At the booth, the wheeled Siasun Robot & Automation Gino 1, unmanned picking station, and several mobile Siasun Robot & Automations were working with high precision coordination, quietly picking out chips, bread, rubber bowls from the bin one by one and accurately placing them into the corresponding seeding wall positions. There were no extra "performances," only a complete end-to-end operation process running smoothly - sorting, handling, delivering, where each link was as smooth and silent as in a real warehouse or factory. The movements were not dazzling - but what the industrial scene required was stability, not dazzle. In displaying the multi-SKU generalization capabilities and end-to-end operational efficiency in a real working scenario, out of over 150 exhibitors at this year's WAIC, Extreme Wisdom was the only one. In their view, the core ambition of Extreme Wisdom is not just to create better Siasun Robot & Automation, but to be the "solution expert" of the embodied intelligence era - based on over a decade of real-world scenario accumulation, providing practical, replicable, and scalable embodied intelligence solutions for various industries. This is a strategic leap from "champion products" to "expert services," and is the central clue to understanding the long-term value of Extreme Wisdom. Gravity Framework: "Operating System of Embodied Intelligence" In 2006, when Amazon AWS quietly launched, no one thought it was a big deal. At that time, everyone's attention was focused on Google, YouTube - those truly "world-changing" dazzling applications. Looking back twenty years later, AWS's market value exceeded the total value of almost all star applications of that era. Now, the wave of embodied intelligence is sweeping in at a stunning speed. Many enterprises are eager to showcase the cutting-edge human-shaped Siasun Robot & Automation Demos, but Extreme Wisdom is going against the grain, focusing on "real machine action," showcasing the stable performance of their wheeled Siasun Robot & Automation Gino 1 in complex, multi-SKU, and generalized scenarios. Behind this "plainness" is a profound strategic judgment: the future of embodied intelligence does not depend on the dazzling performance of individual Siasun Robot & Automation, but on whether they can build a unified, reliable, and open "operating system." The core release of Extreme Wisdom at this WAIC is a unified embodied intelligence framework called Gravity. Gravity adopts a "dual-brain synergy" architecture: the cognitive brain is responsible for understanding instructions, breaking down tasks, while the action brain conducts a physical "sandbox simulation" before executing actions, conceptualizing outcomes, and generating the optimal action sequence. On the opening day of WAIC, Chen Chao, the head of the Large Model Team of Extreme Wisdom, pointed out the core contradiction of the industry with a precise statement: "Models that understand language semantics cannot predict physical movements, and models that can predict physical movements lack high-level semantic understanding." Even more critical is the technological route choice of Gravity 4D. It transcends the limitations of "only replicating images," by strengthening the three-dimensional space and motion 4D physical representation, balancing reasoning speed with cross-materials generalization, utilizing full-body parallel coordinated control, moving and synchronizing grip posture adjustments, and improving task efficiency by about 30%. Data validation: The LIBERO-Plus benchmark's zero-sample success rate has increased from 73.73% to 78.62%, with the most significant enhancement occurring in scenarios of "appearance change but physical invariant" such as camera perspective changes, sensor noise, and lighting changes. GINO ECO Ecology: "Standard Builder and Preacher" The iteration of technology inevitably requires ecological support. Extreme Wisdom's strategy of "one core, two engines" launched now combines the Gravity framework (core) with data engines and ecological engines (two engines), outlining their platform ambitions. If OpenAI's barrier is Internet text data, Extreme Wisdom's barrier is the interaction data of the physical world, and the open ecological engine. This is not a gap that can be quickly caught up with through financing and hiring. Extreme Wisdom's real business network, accumulated over 11 years - covering over 40 countries, with over 1700 projects, serving more than 950 international renowned brands (Walmart, Adidas, Siemens, BMW, etc.) - generates millions of real SKU capture data every day, forming an almost insurmountable data moat that is continually expanding. Extreme Wisdom has also simultaneously launched the GINO ECO Open Ecology: opening the hardware platform to algorithm companies, exploring innovative scenarios with leading clients, and collaborating with complementary hardware companies to create end-to-end workflows. Many people see this as a "win-win stance." But to understand its deeper logic, one needs to refer to another historical case: Android. Google opened up Android to all mobile phone manufacturers, seemingly relinquishing hardware control, but in reality, trading openness for control over access to the mobile Internet, ultimately fostering the prosperity of the entire Android ecosystem. Extreme Wisdom's logic is similar. The more GINO ECO is open, and the more algorithm companies and hardware enterprises are connected, the more difficult it will be to shake the position of the Gravity framework as the "interface standard." As Extreme Wisdom's founder and CEO Zheng Yong said, "Embodied intelligence is a marathon, not a sprint. We are willing to be the super connector of the entire industry." By connecting all aspects, Extreme Wisdom is building a panoramic view of embodied intelligence with Gravity as the core, data as the fuel, and ecology as the branches. Cognitive difference is the real pricing opportunity Behind the strategic ambitions is a business logic that has already been proven. In the 2025 fiscal year, Extreme Wisdom delivered a report that crossed the profit inflection point: revenue of 3.171 billion RMB, a year-on-year growth of 31.6%; adjusted net profit of 43.82 million RMB, achieving a turnaround from loss to profit; and significantly positive operating cash flow. This is a company that has crossed the profit inflection point, and not one that is still burning money to tell stories. At the same time, the management has repurchased 16.2 million shares in over 15 consecutive trading days since June 25 - the sign is self-evident. The explosion of the embodied intelligence market has become a consensus in the industry. Goldman Sachs predicts that by 2035, the global market for human-shaped Siasun Robot & Automation will reach 38 billion US dollars, while Citibank is more optimistic, expecting the global market for human-shaped Siasun Robot & Automation to reach 154 billion US dollars by 2035. Thanks to the industry's momentum and the company's fundamentals, among the 8 brokerage firms that cover Extreme Wisdom, all have given buy or hold ratings. CITIC SEC has the highest target price of 53 Hong Kong dollars, with a comprehensive average target price of about 24.6 Hong Kong dollars - representing an upside potential of over 120% compared to the current stock price. However, more important than the price difference is the switch in cognitive frameworks. The market currently values Extreme Wisdom as a "warehouse AMR hardware company." However, Extreme Wisdom is transforming into a "physical AI operating system" company. The valuation gap between these two frameworks is not just a 20% or 30% correction space but could involve a fundamental restructuring of pricing logic - just as the market revalued NVIDIA from a "game graphics card company" to an "AI computing infrastructure" company.