簡介
Market Definition and Introduction
The Global Robotics Intelligence market was valued at USD 6.24 billion in 2025, and is projected to reach USD 168.37 billion by 2035, growing at a CAGR of 39.03% from 2026 to 2035. Software components lead the market, with foundation model-based intelligence platforms capturing the highest growth rates. Manufacturing and logistics applications together account for the largest combined application revenue share. North America leads with approximately 38% of global revenue, whilst Asia-Pacific is growing fastest. Jensen Huang declared at CES 2025 that physical AI has reached its ChatGPT moment. That's not hyperbole. It's a procurement signal.
Key Market Trends and Analysis
• The global robotics intelligence market was valued at USD 6.24 billion in 2025, growing at a CAGR of 39.03% through 2035.
• NVIDIA at GTC 2026 partnered with ABB, FANUC, Figure AI, KUKA, and Universal Robots to deploy physical AI at production scale globally.
• Physical Intelligence released the embodied AI foundation model pi0.5 in 2025, enabling robots to perform tasks in new environments without prior site training.
• Robot-related startups raised USD 6.4 billion in the first 11 months of 2024 alone, confirming that investor confidence in robotics intelligence has crossed a structural threshold.
• Figure AI secured USD 1 billion in Series C financing in 2025 at a USD 39 billion post-money valuation, backed by NVIDIA, Intel Capital, and Qualcomm Ventures.
• In January 2026, Boston Dynamics and Google DeepMind integrated Gemini Robotics AI with the electric Atlas humanoid for commercial deployment at Hyundai facilities.
• Tesla began producing Optimus Gen 3 humanoid robots in January 2026, confirming automotive OEMs are entering the robotics intelligence commercial deployment phase.
• Foundation model-based learning is the fastest-growing learning model, as VLA architecture submissions at ICLR grew from 1 in 2024 to 164 in 2026.
• Google DeepMind's Gemini Robotics model reduces total robot deployment cost by an estimated 40 to 60% through multi-step task planning grounded in real-world physics.
• Patent filings covering physical AI robotics methods grew 187% in 2024 and 2025 versus the prior two years, confirming accelerating competitive investment in the intelligence layer.
Robotics Intelligence Market Size and Growth Projection
• Market Size in Base Year (2025): USD 6.24 billion
• Market Size in Forecast Year (2035): USD 168.37 billion
• CAGR: 39.03%
• Base Year: 2025
• Forecast Period: 2026–2035
• Historical Data: 2022, 2023, 2024
The Robotics Intelligence market is the AI intelligence layer that enables robots to perceive, reason, learn, and act autonomously in dynamic environments. It covers robotics foundation models, robot reasoning systems, perception software, autonomous decision-making engines, reinforcement learning platforms, navigation intelligence, human-robot interaction systems, and multi-agent coordination platforms. The market explicitly excludes robot hardware, sensors, actuators, and mechanical components. Applications span manufacturing, logistics and warehousing, healthcare, retail, agriculture, aerospace and defence, construction, and hospitality. Deployment is integrated across humanoid robots, industrial robots, warehouse systems, medical robots, agricultural robots, defence robots, and consumer robots. The intelligence layer is what the global robotics industry is buying, and it's the segment generating the fastest compounding procurement commitments globally.
NVIDIA's Jensen Huang said every industrial company will become a robotics company. That prediction has a precise commercial implication: every industrial company will also become a buyer of robotics intelligence software. Boston Dynamics' electric Atlas is already deployed at Hyundai facilities. Agility Robotics' Digit operates in Amazon warehouses. China's NDRC issued directives in June 2024 to promote humanoid development at national scale. The EU AI Act is compelling manufacturers to audit autonomous system decision-making. Each of these forces creates structured procurement that wouldn't exist without them.
In March 2026, NVIDIA announced production-scale robotics intelligence partnerships with ABB, FANUC, Figure AI, KUKA, Universal Robots, and Yaskawa at GTC 2026, deploying Isaac GR00T open models across industrial, humanoid, and surgical robot platforms.
Recent Developments in the Robotics Intelligence Industry
In March 2026, NVIDIA announced a sweeping production-scale physical AI partnership programme at GTC 2026. Partners included ABB Robotics, AGIBOT, Agility, FANUC, Figure, KUKA, Skild AI, Universal Robots, and Yaskawa. NVIDIA unveiled new Cosmos world models, Isaac simulation frameworks, and Isaac GR00T N models. More than 500 robotic developers had already adopted the NVIDIA platform by this point, confirming its position as the dominant robotics intelligence infrastructure provider globally.
In January 2026, Boston Dynamics and Google DeepMind completed the integration of Gemini Robotics AI models with the electric Atlas humanoid robot. The combined platform was deployed to Hyundai and DeepMind facilities, marking the first commercial-scale deployment of a major foundation model within a production humanoid robotics programme. This validates the business case for foundation model-based robotics intelligence at enterprise scale rather than controlled research environments.
In 2025, Physical Intelligence released the embodied AI foundation model pi0.5. This model enables robots to perform tasks in entirely new environments without prior site-specific training. The strategic implication is significant: it removes the cost and time burden of environment-specific robot programming. For logistics and manufacturing buyers, this is the first robotics intelligence capability that genuinely competes with human worker adaptability in unstructured environments.
In January 2025, NVIDIA unveiled the Cosmos world foundation models and the Isaac GR00T N1 model at CES 2025, alongside an energy-efficient Jetson T4000 module powered by Blackwell architecture. Jensen Huang stated that physical AI had reached its ChatGPT moment. LG Electronics, Boston Dynamics, and NEURA Robotics all launched NVIDIA-integrated robots at the same event, confirming the platform's role as the central nervous system of the robotics intelligence commercial ecosystem globally.
Market Dynamics
Humanoid robot commercialisation and labour shortage urgency are driving robotics intelligence market growth.
Barclays projects the humanoid robot market will reach USD 40 billion by 2035, with an optimistic scenario of USD 200 billion. Each humanoid robot requires a continuous intelligence layer for perception, reasoning, and task execution. That software layer is what this market measures. Labour shortages across manufacturing, logistics, and elder care are compelling organisations to deploy intelligent robots, not just automated machines. The difference between automation and robotics intelligence is the ability to adapt to unexpected situations. That adaptation capability is what buyers are now willing to pay premium prices to procure across multiple verticals.
High development complexity and sim-to-real transfer gaps continue restraining robotics intelligence adoption rates.
Building production-grade robotics intelligence systems requires extensive multimodal training data, physics-accurate simulation infrastructure, and large-scale compute investment. Physical AI developers are still solving the sim-to-real gap, where behaviours learned in simulation don't transfer perfectly to physical environments. This gap adds development cost and deployment risk that conservative manufacturing buyers are reluctant to accept without validated performance data from comparable environments. Safety and reliability concerns in unpredictable environments compound the challenge. ABB and FANUC are actively solving this through NVIDIA's Isaac simulation frameworks, but adoption outside early-mover organisations remains constrained by technical maturity thresholds.
Defence robotics procurement and surgical robotics intelligence create substantial new commercial opportunities globally.
Military organisations in the U.S., EU, and Asia-Pacific are deploying autonomous systems requiring advanced perception, navigation, and decision-making intelligence. DARPA's ongoing autonomous systems programme and NATO's robotic combat vehicle requirements are creating structured defence procurement that prioritises reliability and security architecture above cost. Surgical robotics intelligence is equally compelling: CMR Surgical and Medtronic are both building on NVIDIA's physical AI platform, creating a clinical robotics intelligence procurement category that is separate from industrial and logistics applications. These two verticals together create structurally distinct procurement pipelines that don't correlate with manufacturing business cycles.
Paradigm fragmentation and multi-modal perception integration complexity present structural robotics intelligence market challenges.
There is no single convergent architecture for general embodied intelligence models yet. Some companies focus on complex manipulation, others on cross-hardware generalisation, and others on adapting to unstructured environments. For enterprise buyers, this fragmentation creates vendor lock-in risk and integration complexity when deploying intelligence platforms across mixed robot fleets. Multi-modal perception, processing cameras, LiDAR, force sensors, and language simultaneously, demands compute resources and software integration depth that most organisations' robotics engineering teams cannot manage without specialist system integrator support throughout the deployment lifecycle.
Foundation models, world models, and VLA architectures are reshaping the robotics intelligence technology landscape fundamentally.
Vision-Language-Action model submissions at ICLR grew from 1 in 2024 to 164 in 2026. That's the clearest signal available that VLA architectures are the dominant research direction converging toward production deployment. NVIDIA's Cosmos world foundation models, trained to understand physics and spatial relationships from millions of hours of real-world video, enable simulation-based training that transfers to physical environments with dramatically lower data collection costs. The software layer for robotics intelligence is projected to grow at 54.7% CAGR through 2034, making it the market's highest-compounding commercial opportunity for the entire forecast period.
Attractive Opportunities in the Robotics Intelligence Market
• Humanoid Robot Platforms: General-purpose humanoid robots entering production create sustained demand for scalable intelligence platform procurement globally.
• Warehouse Robotics Intelligence: Amazon, DHL, and logistics operators deploying intelligent robots create high-volume, recurring robotics intelligence software procurement globally.
• Defence Autonomous Systems: Military autonomous vehicle and robot programmes create long-cycle, high-specification intelligence platform procurement outside commercial market cycles.
• Surgical Robotics Intelligence: CMR Surgical and Medtronic's NVIDIA-based platforms create structured clinical robotics intelligence procurement across global hospital networks.
• Agricultural Robotics Deployment: Labour shortages in agriculture are driving intelligent robot adoption for harvesting, planting, and monitoring across major farming economies.
• Foundation Model Licensing: Robotics foundation models as commercially licensable AI assets create recurring royalty revenue for developers including Physical Intelligence and NVIDIA globally.
• Simulation Platform Services: NVIDIA Isaac and competing simulation frameworks create managed services procurement for enterprises training robotics intelligence systems at scale.
• Retail Robotics Intelligence: Autonomous in-store robots for inventory, shelf management, and customer service create consistent retail sector robotics intelligence procurement globally.
• Multi-Agent Coordination Systems: Fleet-scale intelligent robot coordination platforms for factories and warehouses create premium enterprise infrastructure procurement globally.
• Construction Robotics Deployment: Labour shortages in construction combined with site safety requirements are creating structured intelligent robot procurement across major infrastructure projects.
Report Segmentation
By Component:
• Software
o Robotics Foundation Models
o Perception Software
o Navigation Software
o Decision Intelligence Platforms
o Motion Planning Software
o Reinforcement Learning Platforms
o Robot Operating Intelligence Platforms
o Multi-Agent Coordination Platforms
• Services
o Integration Services
o Training Services
o Managed Services
o Consulting Services
By Intelligence Type: Perception Intelligence, Navigation Intelligence, Decision Intelligence, Manipulation Intelligence, Collaborative Intelligence, Cognitive Intelligence, Autonomous Intelligence
By Learning Model: Reinforcement Learning, Imitation Learning, Self-Supervised Learning, Supervised Learning, Foundation Model-Based Learning, World Model-Based Learning
By Robot Type: Humanoid Robots, Industrial Robots, Service Robots, Warehouse Robots, Medical Robots, Agricultural Robots, Defence Robots, Consumer Robots
By Application: Manufacturing, Logistics and Warehousing, Healthcare, Retail, Agriculture, Aerospace and Defence, Construction, Hospitality
By End User: Manufacturing Companies, Logistics Providers, Healthcare Organisations, Defence Agencies, Retail Companies, Technology Companies
By Region: North America (U.S., Canada, Mexico), Europe (UK, Germany, France, Spain, Italy, Rest of Europe), Asia-Pacific (China, India, Japan, Australia, South Korea, Rest of Asia-Pacific), LAMEA (Brazil, Argentina, UAE, Saudi Arabia, Africa, Rest of Latin America)
Key Market Players: NVIDIA, Google DeepMind, Microsoft, Amazon, Tesla, Figure AI, Physical Intelligence, Skild AI, Covariant, Sanctuary AI
Report Aspects:
• Base Year: 2025
• Historic Years: 2022, 2023, 2024
• Forecast Period: 2026–2035
• Report Pages: 293
Dominating Segments
Software leads the component segment through foundation model platform adoption and intelligence-as-a-service commercial scale.
Software holds the dominant robotics intelligence component position. Physical AI platform software is projected to grow at 54.7% CAGR through 2034, the highest compounding rate of any robotics segment. NVIDIA's Isaac simulation frameworks and GR00T N models, Physical Intelligence's pi0.5, and Google DeepMind's Gemini Robotics are all software layer products. Robot hardware manufacturers are buying intelligence software rather than building it. ABB, FANUC, Yaskawa, and Universal Robots are all building on NVIDIA's platform rather than developing proprietary intelligence systems from scratch. That's a fundamental shift in how the robotics industry is organised, and it's creating a concentrated software revenue opportunity for the developers who own the intelligence architecture layer globally.
In March 2026, NVIDIA launched Isaac GR00T open models at GTC 2026, with ABB, FANUC, Figure AI, KUKA, and Yaskawa all building production-scale robotics applications on NVIDIA's intelligence software platform.
Foundation model-based learning leads the learning model segment through generalisation capability and cross-task performance advantages.
Foundation model-based learning is the fastest-growing learning model category. VLA architecture submissions at ICLR grew from 1 in 2024 to 164 in 2026, confirming the academic and commercial consensus behind this approach. Google DeepMind's Gemini Robotics model reduces total robot deployment cost by 40 to 60% through multi-step task planning grounded in real-world physics. Physical Intelligence's pi0.5 enables cross-environment generalisation without site-specific retraining. These capabilities make foundation model-based learning commercially superior to reinforcement learning and supervised learning for enterprise buyers who need robots that adapt to changing conditions without reprogramming costs. The market is consolidating around this architecture for production deployment across industrial and logistics applications globally.
In 2025, Physical Intelligence released pi0.5, enabling robots to perform tasks in brand-new environments without prior training. This cross-environment generalisation capability is the commercial breakthrough that enterprise manufacturing buyers have been waiting for.
Humanoid robots lead the robot type segment through platform investment scale and commercial deployment acceleration.
Humanoid robots are the highest-investment robot type category, driven by their potential to perform the widest range of tasks without environment redesign. Figure AI's USD 1 billion Series C at USD 39 billion valuation, Tesla's Optimus Gen 3 production start in January 2026, and Boston Dynamics' Atlas deployment at Hyundai facilities collectively confirm that commercial humanoid procurement has begun. Each humanoid robot requires intelligence systems for manipulation, navigation, reasoning, and human interaction simultaneously. The intelligence requirements per humanoid unit are substantially higher than for single-task industrial robots. That higher intelligence content per unit creates premium average revenue per robot for software platform providers across the forecast period.
In January 2026, Boston Dynamics and Google DeepMind integrated Gemini Robotics with electric Atlas, deploying humanoid robot fleets at Hyundai and DeepMind facilities in the first large-scale commercial humanoid intelligence deployment.
Manufacturing leads the application segment through industrial robot fleet intelligence upgrade procurement scale.
Manufacturing holds the dominant application revenue position. Industrial robot installations globally exceed 3.5 million units, and the intelligence upgrade opportunity across that installed base is enormous. ABB, FANUC, KUKA, Yaskawa, and Universal Robots are all integrating AI intelligence layers into existing robot platforms. Google DeepMind's Gemini Robotics model's 40 to 60% deployment cost reduction is particularly relevant to manufacturing buyers managing large robot fleets where per-unit intelligence integration cost compounds significantly. Patent filings in physical AI robotics methods grew 187% in 2024 and 2025, with the majority originating from manufacturing-focused AI development programmes, confirming manufacturing as the application category attracting the deepest concurrent commercial and research investment globally.
In March 2026, NVIDIA's GTC 2026 partnerships with FANUC and Yaskawa brought Isaac GR00T intelligence models to high-precision electronics assembly and large-scale manufacturing automation programmes, confirming manufacturing as the primary commercial application for production-scale robotics intelligence.
Regional Insights
North America leads the robotics intelligence market through lab concentration and enterprise deployment momentum.
North America commands approximately 38% of global robotics intelligence market revenue. The United States hosts NVIDIA, Google DeepMind, Physical Intelligence, Figure AI, Skild AI, Covariant, Tesla, Amazon, and Microsoft, the firms driving the most commercially significant robotics intelligence developments globally. Amazon's warehouse robotics network, Hyundai's Atlas programme in partnership with Google DeepMind, and defence autonomous systems procurement through DARPA are generating real-world operational data at a scale no other single market currently matches. Federal procurement through the U.S. Department of Defense is a major revenue driver for high-specification physical AI systems requiring secure edge processing and stringent reliability guarantees aligned to military deployment standards.
In March 2026, NVIDIA's GTC 2026 physical AI partnership programme with global industrial robot manufacturers was headquartered in San Jose, confirming North America's continued role as the strategic centre of the global robotics intelligence commercial ecosystem.
Europe accelerates robotics intelligence adoption through industrial automation investment and EU regulatory compliance programmes.
Europe holds a significant robotics intelligence market position, driven by Germany's automotive and industrial manufacturing sectors, the UK's advanced robotics research base, and France's enterprise automation investment. ABB, KUKA, Universal Robots, and CMR Surgical are European companies building on NVIDIA's robotics intelligence infrastructure. The EU AI Act's requirements for autonomous system decision-making auditability are creating compliance-driven procurement for verifiable intelligence platforms across manufacturing and healthcare robot deployments. Germany's automotive sector, including Volkswagen, BMW, and Mercedes-Benz, is actively deploying intelligent robots for body assembly and quality inspection, creating structured enterprise procurement that sustains European robotics intelligence market growth independently of consumer market cycles.
In 2025, CMR Surgical confirmed its Versius surgical robot system was building on NVIDIA's physical AI platform, confirming European medical robotics intelligence as a commercially active procurement category across major hospital networks.
Asia-Pacific drives fastest robotics intelligence growth through China's national humanoid programme and Japan's factory automation investment.
Asia-Pacific is the fastest-growing robotics intelligence regional market. China's National Development and Reform Commission issued directives in June 2024 to promote humanoid development at national scale, creating government-backed procurement that no other single policy programme matches in scale. AGIBOT, Cambricon Technologies, and Huawei's robotics intelligence programmes are building domestic alternatives to NVIDIA and Google DeepMind platforms. Japan's FANUC and Yaskawa are integrating AI intelligence layers across their existing industrial robot installed base. South Korea's Samsung Research and Hyundai, through its Boston Dynamics ownership, are both commercialising humanoid intelligence systems. India's manufacturing automation investment is creating incremental robotics intelligence procurement as Indian factories upgrade to AI-enabled robot fleets.
In January 2025, NVIDIA's GR00T and Cosmos launches at CES attracted over 500 robotic developers to the platform, with Asia-Pacific manufacturers including AGIBOT and LG Electronics among the first to build NVIDIA-integrated commercial robot products.
LAMEA builds robotics intelligence capacity through Gulf industrial automation and defence autonomous systems investment.
The LAMEA region is an accelerating robotics intelligence market, led by Gulf Cooperation Council nations investing in industrial automation under Vision 2030 and equivalent programmes. Saudi Arabia's NEOM project and UAE's industrial diversification initiatives are creating procurement for intelligent robot systems in construction, logistics, and hospitality environments that were previously labour-intensive. Saudi Aramco and ADNOC are both evaluating intelligent robotic systems for oil and gas inspection and maintenance applications, where the safety case for removing human workers from hazardous environments creates a strong economic justification for robotics intelligence investment. Defence autonomous systems procurement across Gulf, Israeli, and Brazilian military programmes is creating a separate structured robotics intelligence procurement stream outside civilian industrial applications throughout the forecast period.
In 2025, Figure AI's commercial expansion programme included Gulf Cooperation Council industrial partners seeking humanoid robot capabilities for logistics and manufacturing applications, confirming LAMEA as an active robotics intelligence procurement market beyond research and evaluation phases.
購物車
LINE
FB