At the 2025 International Consumer Electronics Show (CES) held in Las Vegas,$NVIDIA (NVDA.US)$CEO Jensen Huang's latest speech attracted a lot of attention, but the market was quite surprised that his speech did not focus on the supply and demand of Blackwell or the next-generation AI GPU architecture Rubin as expected. Most of Huang's speech concentrated on introducing NVIDIA's latest technology ecosystem platform in embodied AI, specifically humanoid robots, and development trends, demonstrating that the robot AI large model based on NVIDIA's software and hardware system is becoming increasingly mature and efficient in its training to deployment ecosystem, establishing NVIDIA as the strongest technological base in the field of humanoid robots in the AI era.

NVIDIA (NVDA.US), the tech giant that accurately tapped into one of the largest application markets in the Internet era— the gaming market, is betting that AI humanoid robots will be the largest scale application scenario of the AI era and is also likely to dominate the next industrial revolution, completely transforming human productivity. The 'wave of intelligent machinery'— a period where embodied AI represented by 'AI humanoid robots' is accelerating its infiltration, this long-present technology trend in science fiction appears to be getting closer to us.

Regarding AI humanoid robots, the world's richest person,$Tesla (TSLA.US)$CEO Elon Musk firmly believes that the heavily promoted Optimus robot will be the core driving force behind Tesla's Market Cap growth—he once predicted that the 'Optimus' humanoid robot would be the most essential catalyst for Tesla's Market Cap skyrocketing to 25 trillion dollars, and boldly stated that mass production is expected to be achieved in 2026. At the CES exhibition in early 2025, NVIDIA's leader Jensen Huang repeatedly emphasized in his speech that 'the next frontier of artificial intelligence is physical AI', and he predicted that 'the ChatGPT moment for general humanoid robots is about to arrive'. Before the latest speech, he had mentioned multiple times that one of the largest terminal application scenarios of generative AI, this epoch-making technology, would be AI humanoid robots.

In his CES speech, Huang introduced several of NVIDIA's latest technological achievements in the field of embodied AI, represented by 'AI humanoid robots', including the development platform for the first generative world-based physical model.$COSMOS (0261.MY)$Four brand new Omniverse Blueprints and the high-performance Thor chip applicable to the Robot Concept field, among other technological advancements. Through Omniverse,$Atomix (4625.JP)$(ATOM.CC), Isaac Sim, NVIDIA DGX Cloud, and NVIDIA humanoid robot hardware computing platform, NVIDIA strives to build a hardware and software ecosystem for a robot AI large model based on physical world for comprehensive dynamic simulation and emulation capabilities, leading the development trend in the AI humanoid robot field, and NVIDIA is determined to be the core foundation in the field of 'AI humanoid robots'. The 'cyberpunk world' portrayed in sci-fi movies and games like Blade Runner 2049 and Cyberpunk 2077 seems to be getting closer to us, with AI humanoid robots soon accompanying billions of people on Earth until they grow old.

2024 is hailed by some investment institutions as the 'Year of Awakening for AI Humanoid Robots'. With the rapid development of artificial intelligence technology increasingly integrated into Consumer Electronics application terminals, the AI humanoid robot market could be the next super blue ocean to achieve 'exponential growth'. The recent surge in Tesla's stock price is supported not only by the 'animal spirits' sweeping the market due to Trump's presidential election victory but also by the mass production expectations of Optimus humanoid robots, viewed as key logic by major Wall Street firms like Bank of America. As NVIDIA's humanoid robot hardware and software ecosystem matures, attracting global companies to join the NVIDIA Robot Technology Platform, the market begins to expect that the future robot business may become NVIDIA's next 'explosive growth business' after Datacenter operations. In the A-share market, as market conditions recently weakened, Robot Concept stocks have surged against the trend, driven by intense and ongoing catalysts related to AI humanoid robots, constantly channeling funds into this highly robust investment domain.

Tesla's overseer Musk stated that if everything progresses smoothly, the production of Tesla's Optimus AI humanoid robots will increase tenfold by 2026, amounting to approximately 0.05 million - 0.1 million units. Musk anticipates that this figure will further increase by tenfold to start from 0.5 million units by 2027. A recent Research Report from Goldman Sachs pointed out that humanoid robots are expected to first be applied in global manufacturing factories between 2024 and 2027, while entry into the Consumer market is anticipated between 2028 and 2031; Goldman Sachs predicts that by 2035, global shipments of humanoid robots will reach 1.4 million units, with a market size of approximately 38 billion USD, far exceeding Goldman Sachs' previous forecast of 6 billion USD, whereas the current estimated market size for humanoid robots is even less than 2 billion USD.

The 'Great Wave of Intelligent Machines' is drawing closer to human society! NVIDIA vows to be the 'Pioneer of Intelligent Machines.'

The next wave of artificial intelligence will focus on the humanoid robot field, with one of the most exciting developments being embodied AI. 'We are advancing the entire NVIDIA Robot Technology stack, providing global humanoid robot developers access to technology platforms to use humanoid robot hardware and software technologies, AI acceleration libraries, and specialized AI large models that can significantly enhance the efficiency of robot model R&D.' This is a vision of NVIDIA's humanoid robots revealed by Jensen Huang, known as the 'Godfather of AI,' prior to his speech at CES.

At the 2025 CES exhibition, NVIDIA is doing its utmost to achieve the aforementioned robot vision. They are striving to build an integrated ecosystem of robot AI hardware and software, from blueprint construction to actual AI training, and then to application deployment, based on their "high-performance data center AI GPU system + CUDA ecological barrier" as the underlying technology, especially with the heavy launch of NVIDIA's physical AI system at CES 2025.$Cosmos (ATOM.CC)$The core of the "physical AI system" includes the newly launched Omniverse, the recently upgraded Isaac Sim, and is supported by DGX cloud AI computing resources, dedicated to integrating cutting-edge NVIDIA robotic technology that is based on physical world 3D dynamic simulation and real-world emulation into the robotic model development and training processes of technology companies. It also aims to verify the reliability of robot behavior + vision + foundational models similar to GPT, as well as the integrated logic of the complete robot AI large model in the simulation environment, and their actual capabilities in the real world.

According to NVIDIA's official website, Omniverse is a real-time 3D design collaboration platform that aims to provide developers of real-world simulation and emulation architectures the ability to construct and simulate complex virtual environments. The newly launched Omniverse at CES 2025 stirred waves in the field of robot development, allowing the latest iteration of Omniverse to create high-fidelity simulation environments to assist in the training and testing of robot behavior and visual AI models.

Isaac Sim is a recently launched physics-based robot simulation platform by NVIDIA, aimed at providing an efficient and realistic simulation environment for robot developers and researchers. Based on the Omniverse robot simulation platform, Isaac Sim combines more powerful graphics rendering, a physics simulation engine, and deep learning capabilities to provide high-precision physical simulations and sensor simulations, supporting the development, testing, and verification of intelligent algorithm functionalities such as robot vision and behavior.

The newly launched Cosmos at CES 2025 is a state-of-the-art world foundation model platform capable of generating high-resolution, photo-realistic 3D video streams for training AI humanoid robots and autonomous vehicles, significantly reducing the development costs of large models in these two fields while improving development efficiency. The Cosmos world foundation model platform utilizes over 20 million hours of video training data, aiming to "teach robots to understand the physical world."

At this CES, Jensen Huang stated, "From cars and trucks to factories and warehouses, everything based on mobility will be robotized and driven by AI large models. NVIDIA's Omniverse Digital Twin Operating System and Cosmos physical AI technology are the cornerstones of driving the global real industry towards highly efficient digitalization." Several leading tech companies have already become the first users of Cosmos, including 1X, Agile Robots, Waabi, and others.$Uber Technologies (UBER.US)$ These companies are utilizing NVIDIA's "Physical AI System," which includes Cosmos and Omniverse, to accelerate the development of AI humanoid robots and autonomous driving technology.

At the 2025 CES, the advancement of AI humanoid robots is a significant focus for global tech fans and media. This year, edge AI takes center stage at CES, discussing how AI can better integrate into edge Consumer Electronics and how it seeps deeply into daily human life, especially with the new products and advancements in the crucial field of "AI humanoid robots".

When the deployment scale of Datacenter Server AI Chips and other Hardware meets the benchmark computing power requirements of global enterprises and individuals, as well as the massive computing power support needed for the evolution of AI large models, the trend suggests that similar to OpenAI-o1, AI large models with increasingly powerful AI inference capabilities will undoubtedly be integrated into these devices. They will also harness the cloud's stronger AI computing power for immediate collaboration, which is known as edge AI.

Samsung announced that its AI robot Ballie will begin mass production this year. Ballie is a small rolling AI robot that can interact with users through voice commands and project buttons on the ground for control. It also has the ability to control Smart Home devices based on dynamic environmental changes and security alerts, aiming to become users' edge AI assistant.

Yukai Engineering launched the emotional interactive robot Mirumi at CES, which can simulate infant behavior and interact with users; Enchanted Tools showcased the Mirokai robot, which is intended for future applications in nursing homes and hotel services; Tombot presented a highly realistic emotional companion pet robot, mainly targeted at the elderly or those with health issues who cannot care for real pets, providing genuine emotional companionship and daily assistance.

In December 2024, Boston Dynamics released a new video of its humanoid robot Atlas. In the video, the robot plays Santa Claus, performing movements like kneeling squats, arm swings, flips, and landing steadily. In October 2024, the world's largest Auto Manufacturers, along with Boston Dynamics, the humanoid robot research company under Hyundai Motor, announced a collaboration to accelerate the development of AI large model-based humanoid robots. This move marks that the global auto industry's rightful leader, Toyota Motor, is following Tesla and new Chinese automakers to enter the cutting-edge field of humanoid robotics.$Toyota Motor (TM.US)$

In August 2024, OpenAI, $Microsoft (MSFT.US)$along with many leading figures in the Technology industry such as NVIDIA, and $Amazon (AMZN.US)$the AI humanoid robot startup Figure AI, supported by tech giants like founder Bezos, launched the preliminary version of the second-generation humanoid robot Figure 02. This new humanoid robot by Figure AI fully integrates the multimodal AI large model released by OpenAI, and the Technology sector widely considers this humanoid robot as the "walking ChatGPT," possessing powerful abilities such as mimicking human behavior, deep learning and thinking, and engaging in natural and efficient communication with humans, combining the flexibility of the human form with OpenAI's state-of-the-art AI large model.

Musk boldly claims that the production capacity of Optimus will increase tenfold! Wall Street looks forward to the "mass production moment" of AI humanoid robots.

At the CES exhibition on January 9, Musk revealed key clues regarding Tesla's "AI+" layout involving the Optimus robot and FSD through a video live broadcast. Musk stated, "Humanoid robots will be the greatest product ever, everyone will want at least one robot... Unless someone has hidden some advanced technology, I dare say we have the most advanced and refined robots in the world, especially with regard to the flexibility, precision of robotic hands, which are virtually indistinguishable from human hands."

Regarding the production capacity plan for Optimus, Musk also provided specific targets: Tesla plans to produce thousands of Optimus humanoid robots by 2025, and initially, these robots will undergo production work testing within Tesla factories. "If everything goes smoothly, we plan to increase production to ten times that of 2025 next year (2026), which is approximately 0.05 million to 0.1 million units; the following year, we will further increase it to ten times that, which is starting from 0.5 million units." In short, if Musk's "cake" can land smoothly, the production of Optimus in 2027 will be 100 times that of 2025.

The relationship between Tesla, led by Musk, and NVIDIA is extremely close. Musk's bold prediction may stem from the fact that Tesla, as an 'internal testing player,' has already experienced the NVIDIA robot AI large model ecosystem, which will drive the integration of physical simulation, visual models, and powerful comprehensive capabilities akin to ChatGPT into the accelerated development of robot models. Once the 'brain' of the robot—namely, the robot AI large model system—and the subsequent iterations of the hardware and software infrastructure are completed, mass production will be a breeze for Tesla, which masters the most cutting-edge manufacturing industry.

According to the latest research by TrendForce, as humanoid robots advance toward high-level system integration and are expected to transition from industrial scenarios to household life, the training of AI large models at the front end will become more critical to meet the increasing back-end understanding and interaction needs. The global robot AI large language model (robot AI LLM) market, including AI training and AIGC solutions, is expected to exceed $100 billion by 2028, with a compound annual growth rate of 48.2% from 2025 to 2028. TrendForce indicated that NVIDIA's launch of the Cosmos platform, which contains the World Foundation Model (WFM), at CES 2025 could become an important market momentum.

A recent forecast report released by Markets And Markets indicates that the global humanoid robot market is expected to be approximately $2 billion in 2024, with rapid expansion to $13.25 billion by 2029 and a compound annual growth rate likely exceeding 45%. The research agency stated that starting in 2024, humanoid robots will widely penetrate various industries such as Medical Care and surgical assistance, Auto Manufacturers, and high-risk manufacturing sectors, promising a significant enhancement of operational efficiency across these industries.

Last June, Musk made astonishing remarks at the Tesla shareholder meeting, stating that the 'Optimus' humanoid robot could become the core catalyst for elevating Tesla's Market Cap to $25 trillion. Recently, Bank of America noted in a research report that Tesla is capable of balancing the expansion of its electric vehicle business with bold investments in autonomous driving and humanoid robots. The bank also mentioned that with the deployment of more Optimus robots, Tesla's mass production after 2026 could significantly reduce costs, while the mass production of Optimus will also open up new earnings growth drivers for Tesla.

The underlying technology of 'high-performance datacenter AI GPU systems + CUDA ecological barriers' builds an integrated ecosystem for robot hardware and software, from constructing blueprints to actual AI training, to terminal application deployment. Through the newly launched RTX series ultra-high performance graphics cards at CES, the AI business is extending fully to the personal consumer sector, which is the core logic that leads top Wall Street investment institutions like Wedbush and Rosenblatt Securities to bullishly anticipate NVIDIA's Market Cap reaching $5 trillion in the next 1-2 years. Currently, NVIDIA's Market Cap hovers around $3.43 trillion, only second to Apple, making it the second-largest Market Cap company globally.

Wedbush recently stated in a report: 'The most important message conveyed by Jensen Huang is that NVIDIA will launch a series of new AI technologies in the fields of humanoid robots, autonomous driving, and personal computers, which will further expand NVIDIA's significant hardware and software integrated technological advantage compared to other semiconductor companies and large technology firms.' Wedbush indicated that the humanoid robot and autonomous driving market represents another incremental market opportunity valued at at least $1 trillion, potentially driving the company's market valuation over $4 trillion in the next 12 to 18 months, and it may even reach $5 trillion.

Editor/ping