We maintain our global humanoid robot shipment forecasts of 51,000 units in 2026E and 76,000 units in 2027E. This represents a multi-fold increase from the estimated 15,000-20,000 units in 2025, primarily driven by dedicated-purpose commercial deployments ahead in addition to entertaining stage performances, scientific study, education and data factory demand. The market is likely to react positively to key humanoid robot supply chain stocks for a few trading days, anticipating an adoption surge in the comi ...

Core Insights - The article presents a counter-consensus viewpoint from Fei-Fei Li, emphasizing that large language models alone cannot lead to Artificial General Intelligence (AGI), and that spatial intelligence is a more foundational path [4][5][6]. Group 1: AGI Route Debate - Language is not the entirety of intelligence and is not its foundation; spatial intelligence, which has evolved over 500 million years, is crucial for AI development [5][6]. - If AI only possesses language capabilities, it will remain confined to the digital realm; true AGI requires understanding and interaction with the three-dimensional physical world [6]. Group 2: Redefining World Models - The newly introduced spatial intelligence model, Marble, can process multimodal inputs and create a navigable, interactive 3D world with physical consistency, differing from traditional video models [7][8]. - Marble has applications in various fields, including game development, visual effects, and even therapeutic settings for conditions like OCD [8]. Group 3: Scaling Law and Data Challenges - The slower development of physical world AI compared to language models is attributed to the noise in physical data and the difficulty in large-scale data acquisition [8][9]. - World Labs employs a hybrid data strategy, combining existing internet data with synthetic and real-world data to overcome these challenges [8][9]. Group 4: General Robotics vs. Autonomous Driving - General robotics is viewed as a higher-dimensional challenge compared to autonomous driving, which operates primarily in a 2D space [10][11]. - The core task of general robots involves interaction in 3D space, which presents significant technical challenges [10][11]. Group 5: AI as a Fundamental Infrastructure - AI is likened to electricity, with its success not measured by model size but by its ability to empower civilization and improve individual lives [11][12]. - The goal of World Labs is to integrate spatial intelligence into various industries, aiming for significant advancements by 2026 [12].

Core Insights - Waymo is enhancing its self-driving technology through the development of the Waymo World Model, which is based on Google DeepMind's Genie 3, aimed at improving real-world service scalability [1][2] Group 1: Waymo World Model - The Waymo World Model utilizes Genie 3's extensive world knowledge to simulate various scenarios, including extreme weather and safety-critical events [3][4] - This model allows engineers to modify simulations using simple language prompts and driving inputs, enhancing the controllability of the simulations [3][4] Group 2: Impact of Genie 3 - Genie 3 is designed to create 3D environments governed by physics, enabling AI agents to learn through exploration of virtual worlds rather than relying on static datasets [5] - Google DeepMind launched an experimental prototype, Project Genie, which allows users to interact with world-generation features [6] Group 3: Market Reaction and Investment - Following the announcement of Genie 3, the video game industry experienced a significant market value loss due to concerns over AI's capability to generate video games [7] - Waymo successfully raised $16 billion in a funding round, resulting in a post-money valuation of $126 billion, with Alphabet remaining its majority investor [7]

Core Viewpoint - Manifold AI, founded by a former key member of SenseTime, aims to redefine embodied intelligence through its World Model technology, enabling robots to not only perceive but also predict physical interactions in their environment [1][4][12]. Group 1: Financing and Growth - Manifold AI has completed over 300 million yuan in financing within just seven months of its establishment, showcasing a rapid fundraising pace that reflects strong market interest in "Physical AI" [2][7]. - The company has successfully raised funds in three rounds: a seed round led by Inno Angel Fund, followed by two angel rounds, each exceeding 100 million yuan [4][7]. - The latest funding round included notable investors such as Meihua Venture Capital, Junlian Capital, and Huawei Hubble, indicating a strong backing from the industry [1][9]. Group 2: Technology Development - Manifold AI's technology focuses on World Model Action (WMA), which allows robots to predict physical state changes based on first-person perspective videos, moving beyond traditional visual-language models (VLM) [12][14]. - The company's WorldScape model enables robots to simulate and interact with their environment autonomously, marking a shift from mere execution of pre-set codes to possessing "brain-like" capabilities [14][15]. - Manifold AI is developing multiple specialized models, including DriveScape for autonomous driving, RoboScape for physical interaction, and AirScape for drones, all built on the foundational WorldScape model [15]. Group 3: Future Aspirations - The company aims to equip over 10% of robots in the market with its "Manifold Brain," pushing the boundaries of Physical AI agents [19][20]. - The long-term vision includes transitioning World Models from experimental stages to practical applications in warehouses, factories, and homes within the next three years [20][21]. - The strategy emphasizes creating a universal embodied world model while simultaneously commercializing sub-domain models to generate revenue and support further development [20].

Core Viewpoint - The autonomous driving industry has entered its second phase, requiring more dedicated individuals to address its challenges and pain points [2]. Group 1: Industry Direction - The main focus areas include but are not limited to: autonomous driving product management, 4D annotation/data loop, world models, VLA, large models for autonomous driving, reinforcement learning, and end-to-end solutions [4]. Group 2: Job Description - The positions are primarily aimed at training collaborations in autonomous driving, targeting B-end (enterprises, universities, research institutes) and C-end (students, job seekers) for course development and original article creation [5]. Group 3: Contact Information - For discussions regarding compensation and collaboration methods, interested parties are encouraged to add the WeChat contact wenyirumo for further communication [6].

Core Insights - The article presents 20 key predictions for AI trends in 2026, highlighting significant advancements and shifts in the industry [2] Group 1: AI Paradigms and Trends - The emergence of a new paradigm in AI, focusing on continual learning, is expected to gain traction in 2026, with positive signals likely to emerge from at least 1-2 technical pathways [5] - ChatGPT is projected to double its daily active users (DAU) to between 800 million and 1 billion by 2026, establishing itself as a global entry point for users [6] - The "App-store Moment" for ChatGPT is anticipated, leading to the creation of the first application generating $100 million ARR within its ecosystem [7] Group 2: Company Developments and Market Dynamics - OpenAI is expected to reverse its narrative in the second half of 2026, potentially achieving a valuation exceeding $1 trillion due to its strong market position and partnerships [9] - xAI's integration into Tesla is predicted to enhance the synergy between digital and physical worlds, contributing to advancements in AGI [11] - 2026 is forecasted to be a significant year for Enterprise AI, with Anthropic's ARR expected to at least double, reaching over $20 billion [12][14] Group 3: Technological Innovations - The multi-modal AI sector is anticipated to experience a commercial breakthrough, with the emergence of applications akin to Pokémon GO [15][16] - Long-horizon tasks and multi-modal demands are expected to drive the growth of new data companies, each achieving $1 billion ARR [17] - Personalization is projected to become a key competitive advantage for leading AI models, enhancing user engagement [19] Group 4: Market Valuations and IPOs - The AI IPO market is expected to flourish in 2026, with significant companies like SpaceX and OpenAI planning to go public, potentially signaling a peak in market sentiment [32] - Google is predicted to surpass a market valuation of $5 trillion, driven by its strong position in the AI model landscape and advertising business [34] Group 5: Infrastructure and Hardware - Nvidia's aggressive investment in optical interconnect technology is expected to lead to a wave of mergers and acquisitions in the CPO sector [27][28] - The demand for storage solutions is projected to surge due to the multi-modal revolution, integrating storage deeply into computational cores [29] - A significant increase in reasoning power is anticipated, with token consumption expected to grow by at least 10 times in 2026 [30][31]

Core Viewpoint - The article discusses the emergence of World Models in AI, emphasizing the importance of Masking as a foundational principle for building these models, which are seen as essential for achieving Artificial General Intelligence (AGI) [1][3][5]. Group 1: Definition and Components of World Models - The true World Model is defined as an organic system composed of three core subsystems: a Generative Heart, an Interactive Loop, and a Memory System [6][8]. - The Generative Heart ($G$) predicts future states and simulates world dynamics, while the Interactive Loop ($F,C$) allows for real-time interaction and decision-making [8]. - The Memory System ($M$) ensures continuity over time, preventing the world from becoming a series of fragmented experiences [8][9]. Group 2: Evolution of World Models - The evolution of World Models is categorized into five stages, with Masking being the central theme throughout these stages [10][12]. - Stage I focuses on Mask-based Models, highlighting Masking as a universal generative principle rather than just a pre-training technique [13][24]. - Stage II aims for Unified Models that process and generate all modalities under a single architecture, with a debate between Language-Prior and Visual-Prior modeling approaches [25][26]. Group 3: Interactive Generative Models - Stage III introduces Interactive Generative Models, where models respond to user actions, transforming from mere simulators to interactive environments [36][40]. - The Genie series, particularly Genie-3, represents the state-of-the-art in real-time interactive models, achieving 720p resolution and 24fps frame rates [41][42]. Group 4: Memory and Consistency - Stage IV addresses Memory & Consistency, focusing on the need for persistent memory to prevent catastrophic forgetting and state drift in generated worlds [46][48]. - Solutions proposed include Externalized Memory, architecture-level persistence, and consistency governance to maintain coherence in generated environments [49][50]. Group 5: Ultimate Form of World Models - Stage V envisions True World Models that exhibit persistence, agency, and emergence, allowing for complex interactions and societal dynamics within the simulated world [51][52]. - The article concludes with the challenges of coherence, compression, and alignment that must be addressed to realize these advanced models [58].

Core Insights - The rapid development of AI large models is becoming a key factor in the new round of technological competition, with a belief that the number of foundational large models will converge to a single-digit figure, while numerous specialized models and applications will emerge across various industries [1] - Physical AI is highlighted as a significant area of focus, accelerating advancements in embodied intelligence and autonomous driving, which are expected to profoundly change societal operations [1] - The transition to the "Agent era" presents challenges in integrating AI technology into the real economy, particularly in terms of legal, compliance, and ethical considerations [1] Physical AI Debate - The emergence of Sora in early 2025 has sparked discussions about "world models" and the competition between two core routes of physical AI: world models and VLA (Visual Language Models) [2] - Sora's development signifies AI's evolution from a "predictor" to a "simulator," marking a paradigm shift necessary for applications like autonomous driving and embodied intelligence [2] - Current models like Sora are criticized for being mere "visual simulators" lacking true physical world modeling capabilities, as they often fail to maintain physical logic [2][3] Model Differentiation - The world model route has diverged into "generative" and "representational" factions, with generative models like Sora focusing on empirical learning from vast sensory data, while representational models emphasize rational deduction through structured internal representations [3] - Generative models are suited for data factories or simulation training, whereas representational models excel in decision-making processes [3] Industry Trends - There is a trend towards the integration of VLA and world models, utilizing VLA for high-level strategy planning and world models for low-level action validation [4] - The evolution of network architecture is shifting from "cloud-native" to "AI-native," necessitating networks to achieve extreme performance and seamless integration of computing and networking [5][6] AI Native Applications - AI applications are transitioning from content generation to autonomous action, with a focus on restructuring entire value chains rather than merely enhancing efficiency in isolated processes [7] - The challenges of deploying agents in critical industries like telecommunications and finance include reconciling the randomness of models with deterministic business needs and ensuring stability in long-term tasks [8] Deep Water Practices - Industries that are likely to achieve scalable AI value realization include education, healthcare, software development, intelligent manufacturing, and urban governance, characterized by high data structuring and rapid feedback mechanisms [9][11] - The transition from "shallow water" to "deep water" signifies AI's deeper integration into core business processes, facing complexities such as multi-modal data and new security threats [12] Hybrid Approaches - The development paths for AI integration may involve a hybrid approach combining "general foundational models + industry fine-tuning" and building industry-specific small models from scratch [12][13] - General models trained on human language may introduce noise in industrial applications, necessitating the creation of specialized models for non-natural language data [13]

Group 1 - The core viewpoint of the article highlights the competitive landscape of the autonomous driving industry, emphasizing the focus on technology, cost, and efficiency as key areas of competition this year [1] - The industry has seen a shift with many professionals transitioning to sectors like embodied AI and drones, while autonomous driving remains a mature AI field, making algorithm talents highly sought after [1][2] - Major technological directions in autonomous driving have converged this year, including end-to-end systems, VLA, world models, and reinforcement learning, with many midstream companies tackling challenges like OCC and multi-sensor fusion perception [3] Group 2 - The membership of the paid community focused on autonomous driving has officially surpassed 4,000, indicating a growing interest in the development of technology routes and job information [3] - The company expresses gratitude to its supporters and announces various benefits and discounts for the new year, encouraging continued efforts in the upcoming year [4]

Core Insights - The autonomous driving industry has experienced significant developments this year, focusing on technology, cost, and efficiency improvements as it matures [1] - There has been a notable shift in talent, with many professionals transitioning to other sectors like L4, embodiment, and drones, while algorithm talent in autonomous driving remains highly sought after [1][2] - Major technological advancements in autonomous driving have consolidated around key areas such as end-to-end systems, VLA, world models, and reinforcement learning, with many midstream companies actively hiring [3] Industry Trends - The autonomous driving sector is seeing an increase in B-end clients and a movement towards offline engagement, while C-end services are becoming more specialized [1] - The community of paid members in the autonomous driving sector has surpassed 4,000, indicating growing interest and engagement in technology development and job opportunities [3] - The industry is characterized by strong collaboration capabilities among professionals who have experience with large clusters and corner cases, which are lacking in other sectors [2]