Financial Data and Key Metrics Changes - Revenue grew 26% year-on-year to a record $1.24 billion, marking the fourth consecutive billion-dollar quarter [5][13] - Royalties increased 27% to a record $737 million, driven by strength in AI and general-purpose data centers [5][13] - Non-GAAP EPS reached $0.43, supported by higher revenue and slightly lower operating expenses than expected [16] Business Line Data and Key Metrics Changes - License revenue was $505 million, up 25% year-on-year, driven by demand for next-generation technologies [5][14] - Data center royalty revenue has grown more than 100% year-on-year, with expectations for it to become the largest business segment in the future [5][13] - Edge AI devices, particularly smartphones, are experiencing faster growth than the market, with all major Android OEMs ramping up production of CSS-based chips [13][14] Market Data and Key Metrics Changes - Arm's share among top hyperscalers is expected to reach 50%, with significant deployments of Neoverse CPUs [8][9] - The automotive market in Physical AI grew double digits year-on-year, contributing to strong royalty performance [14] - The shift towards agent-based AI is reshaping data center design, requiring CPUs with higher core counts and better power efficiency [8][10] Company Strategy and Development Direction - Arm has organized its business around three units: Edge AI, Physical AI, and Cloud AI, to align with customer deployment of AI [6] - The company is focused on investing in innovation across a broad spectrum of compute technologies, including next-generation architectures and compute subsystems [5][16] - Arm aims to be the compute platform of choice for all AI workloads, leveraging its strengths in power efficiency and predictable latency [10][91] Management's Comments on Operating Environment and Future Outlook - Management expressed confidence in future revenue growth due to strong customer demand and a growing base of long-duration contracts at higher royalty rates [17] - The company anticipates revenue of $1.47 billion for Q4, representing an 18% year-on-year growth at the midpoint [17] - Management acknowledged potential risks from memory supply chain constraints but indicated that growth in Cloud AI is compensating for these risks [24][25] Other Important Information - Arm is hosting an event on March 24th, with no details provided ahead of the event [18] - The company is exploring chiplets and complete SoCs as part of its R&D investments [16] Q&A Session Summary Question: Arm's role in AI and cloud data centers - Management highlighted the shift from training to inference workloads, emphasizing the suitability of CPUs for agentic AI tasks due to their power efficiency and low latency [21][22] Question: Impact of memory supply chain constraints on royalty revenue - Management indicated that a potential 20% reduction in smartphone unit volumes could translate to a 1-2% negative impact on total royalties, with Cloud AI growth offsetting risks [23][24][25] Question: SoftBank's potential need to sell Arm stock - Management confirmed that SoftBank's leadership is not interested in selling any shares of Arm stock, expressing long-term confidence in the company [30] Question: Trends in royalty revenue growth - Management noted that royalty growth percentages may be lower due to tougher comparisons from previous quarters, but absolute dollar growth is expected to remain strong [31][32] Question: Data center revenue quantification - Management indicated that data center revenue is expected to grow significantly, potentially reaching similar or larger levels than the smartphone business in the coming years [39] Question: Impact of higher royalty rates on smartphone unit volumes - Management explained that the transition to higher royalty rates with v9 and CSS will help offset lower smartphone unit volumes [42][43] Question: Partnerships and custom ASICs with SoftBank - Management did not provide specific details on potential custom ASICs but acknowledged the substantial partnership with SoftBank [46] Question: Arm's IP penetration in AI data center semis - Management discussed the evolving architecture of data center chips and the increasing role of CPUs in handling AI workloads [49][50] Question: Compute subsystems' contribution to royalty revenue - Management indicated that CSS has grown from approaching double digits to well into the teens percentage of royalty revenue, with expectations for further growth [56][57]

Core Insights - GSI Technology announced preliminary benchmark results for its Gemini-II Compute-in-Memory processor, achieving a time-to-first-token (TTFT) of 3 seconds for multimodal large language models at the edge, with a power consumption of approximately 30 watts [1][2]. Performance Metrics - The Gemini-II processor demonstrated a TTFT of 3 seconds, which is the lowest reported for a multimodal 12B model on an embedded edge processor [2]. - Competitive platforms reported TTFTs of approximately 12 seconds on Qualcomm Snapdragon X Elite at 30W and 3 seconds on NVIDIA Jetson Thor at over 100W, indicating that Gemini-II offers superior performance at lower power levels [3]. Market Implications - The performance profile of Gemini-II is well-suited for "physical AI" markets, including drones and smart city applications, where power and thermal constraints are critical [4]. - The shift from cloud-assisted models to local inference in edge physical AI is expected to enhance latency, reliability, and operational efficiency [5]. Development and Collaboration - GSI's engineering team is focused on optimizing the responsiveness of the Gemini-II processor while collaborating with partners like G2 Tech for system integration and proof-of-concept activities [6].

Core Insights - Manifold AI has completed over 100 million yuan in angel+ round financing, with Jinqiu Capital continuing to invest. The funding will be used for the iteration of its world model and the application of embodied intelligence [4] - The company has developed a universal spatial world model called WorldScape, which matches the quality and real-time capabilities of leading global models [6] - Manifold AI is the first team globally to deploy a comprehensive outdoor, indoor, and aerial embodied world model, significantly enhancing data efficiency and model performance [9] Financing and Investment - The latest funding round was led by Junlian Capital, with participation from Meihua Venture Capital, Huawei Hubble, and existing investors including Inno Fund and Jinqiu Capital [4] - Manifold AI has raised several hundred million yuan in total funding over the past six months [4] Technological Advancements - WorldScape enables single-image generation of interactive spaces, providing a foundation for physical AI applications [8] - The company utilizes a vast amount of physical video data for pre-training, enhancing WorldScape's operational interaction capabilities [8] - Manifold AI's approach replaces traditional VLM models with its world model, resulting in superior performance in real-world applications [10] Future Prospects - The integration of NVIDIA Jetson Thor for deploying embodied world models is a significant step towards scaling operations [14] - The involvement of Huawei Hubble is expected to facilitate the integration of domestic chips and robotic brains, laying the groundwork for large-scale implementation [14]

Group 1 - The core product of the company, the Gemini330 series 3D camera, has been successfully adapted to the NVIDIA Jetson Thor platform, enhancing the capabilities of robotic manufacturers by enabling high-speed processing and flexible integration options [1][2] - The collaboration between the Gemini series and the Jetson Thor platform provides a complete visual computing solution for embodied intelligence applications, accelerating innovation in humanoid robots and advanced autonomous mobile robots (AMR) [2] - The company has established a dual manufacturing system in Shunde, China, and Vietnam, which supports global clients with integrated R&D and manufacturing capabilities [3][5] Group 2 - The Shunde manufacturing base has the capacity for millions of sensors and hundreds of thousands of robotic terminals, while the Vietnam factory is expected to commence production in May 2026, enhancing global delivery efficiency and supply chain resilience [5] - The company has successfully provided manufacturing services for leading global clients in various sectors, including rehabilitation robots, smart lawn mowers, humanoid robots, and warehouse and cleaning robots, helping to reduce overall costs and shorten time-to-market [5]

Core Insights - The "ChatGPT moment" for Physical AI has arrived, marking a significant shift of AI technology from virtual screens to the physical world, leading to a transformative phase in the robotics industry [1][2] Group 1: Physical AI Development - Huang emphasized the four stages of AI evolution: Perception, Generation, Agentic, and Physical AI, with the latter enabling models to understand real-world physical laws [2] - The introduction of three core open-source models aims to lower the development barrier for Physical AI, creating a closed-loop technology from environmental cognition to action execution [2][4] - The NVIDIA Cosmos Transfer 2.5 and Cosmos Predict 2.5 models can generate synthetic data that adheres to physical laws, providing a safe virtual testing environment for developers [2] Group 2: Cognitive Reasoning and Robotics - The NVIDIA Cosmos Reason 2 visual language model enhances machines' human-like visual reasoning and decision-making capabilities [3] - The NVIDIA Isaac GR00T N1.6 model achieves a 40% increase in task success rates and reduces training time from three months to 36 hours, improving data efficiency by 60 times [3] Group 3: Open-source Ecosystem - NVIDIA's collaboration with Hugging Face integrates GR00T models and Isaac Lab-Arena into the LeRobot open-source library, connecting 2 million NVIDIA developers with 13 million Hugging Face AI builders [5] - NVIDIA has contributed 650 open-source models and 250 datasets to Hugging Face, leading in resource download volume within the open-source community [5] Group 4: Hardware Upgrades - The new Jetson T4000 module, based on the Blackwell architecture, offers a fourfold performance increase over its predecessor, while the Jetson Thor robot computer is becoming a focal point for industry collaboration [6] - The IGX Thor platform is set to launch, catering to various computational needs in industrial edge scenarios [6] Group 5: Industry Collaboration and Applications - A diverse range of robots, including humanoid, wheeled, and surgical assistance robots, showcased the cross-domain adaptability of Physical AI technology [7] - Major industry players like Franka Robotics and Mercedes-Benz are leveraging NVIDIA's technology to enhance robot training and develop AI-driven products for smart transportation [7]

Core Insights - The Dynamis Robotic Surgical System has received FDA clearance and is currently in routine clinical use in Las Vegas [1] - The system is set to evolve by integrating NVIDIA technologies, including Jetson Thor, Isaac for Healthcare, and Cosmos platforms, to advance hard tissue robotic surgery [1] - LEM Surgical is participating in the 2026 Consumer Electronics Show (CES) to showcase its innovations in next-generation hard tissue robotics [1]

Core Insights - Richtech Robotics has launched Dex, its first mobile humanoid robot designed for industrial applications, in collaboration with NVIDIA to enhance its capabilities [1][3] - Dex utilizes NVIDIA Jetson Thor technology for real-time reasoning and complex task execution, operating efficiently for a full workday on a single charge [2][5] - The robot combines insights from over 450 previous deployments, integrating autonomous mobile robot (AMR) technology with dual-armed precision to enhance operational efficiency [4][5] Technology and Development - Richtech employs NVIDIA's Isaac Sim for training Dex in diverse industrial contexts, facilitating a "Sim2Real" pipeline that accelerates deployment and improves safety [3][6] - The robot's design prioritizes mobility and dexterity, featuring a wheeled platform for stability and lower energy costs, while maintaining a four-hour battery life in mobile mode [4][5] - Richtech is launching an American robotics data initiative to collect regionally grounded data, aiming to empower the development of physical AI in the U.S. [7] Applications and Capabilities - Dex is capable of performing a variety of light to medium industrial tasks, making it a valuable asset in manufacturing and logistics sectors [8][9] - The robot's features include modular end-effectors for various tools, a four-camera vision system for navigation, and the ability to operate continuously from a static base [5][6] - Richtech invites companies to explore pilot opportunities with Dex, showcasing its capabilities at industry events [10][9] Company Overview - Richtech Robotics focuses on developing advanced robotic solutions and data infrastructure, emphasizing automation and continuous AI-driven improvement across various sectors [11]

Core Insights - Advantech has launched a new lineup of Edge AI solutions powered by NVIDIA Jetson Thor modules, setting a new benchmark for edge AI performance with up to 2070 FP4 TFLOPS [1] - The solutions are designed for real-world applications in robotics, medical AI, and data AI, featuring integrated hardware-software platforms and application-specific tools [1][3] Group 1: Edge AI Solutions - The new Edge AI solutions include robotic controllers for humanoid robots, AMRs, and unmanned vehicles, providing real-time AI reasoning and inference with GPU-accelerated SLAM [2] - Advantech's robotic controllers support multi-camera GMSL, 2D/3D sensors, and IMUs, enabling rapid integration and deployment with features like hardware time sync and anti-vibration design [2] Group 2: Medical AI Systems - Advantech's medical AI systems leverage NVIDIA Jetson Thor and advanced SDKs to enhance real-time sensor processing and image analysis for surgical robotics and diagnostics [3] - The platforms aim for low latency and high precision, making them suitable for operating rooms and clinical workflows [3] Group 3: Data AI Systems - The AIR-075 system is designed for data AI applications, featuring powerful computing capabilities with 4× 10GbE and GMSL interfaces [4] - It integrates with NVIDIA AI technologies to enable sensor fusion and real-time predictive edge intelligence [4] Group 4: Container Solutions - Advantech Container Catalog (ACC) offers a range of ready-to-develop edge AI applications optimized for NVIDIA Jetson platforms [5] - The containerized architecture supports scalable edge AI expansion, facilitating development from single-node setups to distributed networks [5]

Core Insights - Jinqiu Fund has completed an investment in Manifold AI, focusing on world models and embodied intelligence, with a total of over 100 million yuan raised in two funding rounds [2][4] - Jinqiu Fund emphasizes a long-term investment philosophy, seeking groundbreaking technologies and innovative business models in the field of general artificial intelligence [3][16] Investment Overview - The recent angel round of financing for Manifold AI was led by Jinqiu Fund, with participation from co-investors including Chuangweiye and existing shareholder Inno Angel Fund [4] - The seed round was led by Inno Angel Fund, with follow-on investment from the Waterwood Tsinghua Alumni Seed Fund [4] Technological Focus - Manifold AI's original embodied world model technology aims to drive the large-scale deployment of robotic brains, addressing the challenges of diverse bodies, limited data, and fragmented applications in general robotics [6][16] - The company utilizes a World Model Action (WMA) approach, leveraging vast amounts of ego-centric video data for pre-training, which is expected to enhance physical space intelligence emergence [10][16] Industry Context - The rapid evolution of robotics and the need for autonomous operational capabilities are critical for large-scale implementation [6] - The shift in technology strategies by companies like Tesla and Figure AI towards using extensive ego-centric video data for training reflects a broader trend in the industry [6][7] Team and Leadership - Manifold AI's core team is based in Beijing, with members having backgrounds in robotics and large models, and experience in developing AI products with millions of users [12] - The founder and CEO, Dr. Wu Wei, has extensive management experience and previously led the development of the world model at SenseTime [13][16] Future Outlook - Jinqiu Fund anticipates exploring the next generation of embodied intelligent world models in collaboration with Manifold AI, as the industry moves towards a deeper understanding of machine interaction with the world [17]

Core Viewpoint - The article highlights the advancements of the "Kua Fu" humanoid robot developed by Leju, which is now equipped with NVIDIA Jetson Thor, enabling it to perform various industrial applications effectively, thus accelerating the commercialization of robotics in multiple sectors [1][3]. Group 1: Technological Advancements - "Kua Fu" features over 40 degrees of freedom in its bionic structure, enhancing its flexibility and adaptability for complex tasks [3]. - NVIDIA Jetson Thor provides 2070 FP4 TFLOPS of AI computing power, significantly improving the robot's operational capabilities across various industrial scenarios [3][8]. - The bandwidth of NVIDIA Jetson Thor has been increased by 35%, allowing "Kua Fu" to maintain stable performance during long-duration tasks [5]. Group 2: Industrial Applications - In logistics, "Kua Fu" demonstrates excellent long-term stability in package sorting applications, effectively handling diverse item shapes and sizes [5]. - In smart manufacturing, the robot's full-body coordination is crucial for SMT tray retrieval tasks, ensuring high precision and compliance with operational standards [8]. - In the 3C electronics sector, "Kua Fu" can achieve rapid recognition and action response within milliseconds, essential for high-speed material sorting [10]. Group 3: Enhanced Capabilities - "Kua Fu" can autonomously adjust its grasping posture to handle various sizes and weights of containers, showcasing its spatial understanding and decision-making abilities [12]. - The robot excels in complex operations such as the precise placement of daily chemical products, requiring advanced tactile and force perception [14][15]. - The integration of NVIDIA Jetson Thor enhances "Kua Fu's" ability to process multi-modal sensor data, improving its agility, decision-making speed, and operational efficiency [15]. Group 4: Future Directions - Leju aims to continue exploring key technologies and innovative applications in humanoid robotics, pushing towards broader industrial and commercial integration [15].