Modern data centers are power intensive, and the growing demand for AI computing creates a challenge for operators to maintain physical infrastructure, manage dense GPU deployments, optimize cooling systems and ensure uninterrupted operations. Robotics and physical AI offers a potential solution by automating repetitive physical tasks, improving operational efficiency, increasing safety and enabling effective facility management. This deployment of automatic management could change data center operations from labour intensive maintenance towards predictive and autonomous operations.
In this special feature, we explore how robotic automation in data centers can perform a variety of tasks to increase efficiency and optimize operations. These include conducting infrastructure inspections, transporting heavy server equipment, monitoring cooling systems, identifying potential equipment failures and operating in hazardous environments. Equipped with technologies such as computer vision, thermal imaging, LiDAR and artificial intelligence which collects operational data, detects anomalies and prevents infrastructure failures.
What makes physical AI exciting?
Physical AI represents a new stage of intelligent data center infrastructure that extends artificial intelligence beyond data processing into the physical machines. By combining AI with robotics, computer vision, sensors, digital twins, and autonomous control systems, machines can perceive and perform physical tasks. This convergence has the potential to transform data center operations, maintenance, security, and energy efficiency while reducing reliance on human intervention.
Key applications include autonomous robots for equipment inspection, fault detection, and hardware maintenance, as well as AI driven cooling systems that optimize airflow and liquid cooling based on real time conditions. Digital twins enable predictive maintenance and infrastructure optimization, while autonomous security systems enhance surveillance, intrusion detection, and personnel safety.
“GenAI amplifies how enterprises think; physical AI transforms how they operate,” Parm Sandhu (Group Vice President: Enterprise 5G Products and Services, NTT DATA), told w.media in a recent exclusive interview. “It moves intelligence off the screen and into machines and operations where the largest enterprise value pools sit: manufacturing, logistics, energy, mining. This is not a hypothetical shift.”
The physical AI sector is attracting significant investment, with major deals and funding rounds indicating a growing demand for computing infrastructure and AI-powered robotics. According to report by Fortune Business Insight, the global physical AI market was valued at US$ 33.42 billion in 2025, and is projected to grow from US$ 42.06 billion in 2026, to US$ 294.07 billion by 2034, registering a CAGR of 27.5 percent during the forecast period.
North America dominated the market in 2025, accounting for a 35.06 percent share. By integrating AI models with sensors, physics-aware reasoning, simulation technologies, and actuators, it enables applications such as robotics, autonomous vehicles, drones, and intelligent machinery to make autonomous decisions, perform physical tasks, and adapt to changing real-world conditions while accounting for physical constraints, time, safety, and uncertainty.
Among recent developments, Nscale and Figure have signed a multi-year partnership with humanoid robotics, committing an initial US$ 3.5 billion in AI computing capacity, with plans to expand the deal to more than US$ 6 billion. The agreement could involve deploying up to 100,000 NVIDIA GPUs to support the development and scaling of Figure’s AI models and humanoid robots, with the initial deployment expected in the second half of 2027.
“Humanoid robots extend physical AI into the world designed for people—opening a major new industry,” said Jensen Huang, founder and CEO, NVIDIA. “Nscale and Figure have activated the robotics flywheel: training Figure’s models on NVIDIA Vera Rubin through Nscale’s AI cloud, validating them in NVIDIA Isaac Sim, and deploying them on NVIDIA GPUs in Figure’s robots. This is the physical AI flywheel that will accelerate the path from models to robots in the world.”
In a separate development, LG Electronics and NVIDIA are planning to establish an AI factory in Seoul to support robotics and industrial AI development. While financial details of the LG-NVIDIA initiative have not been disclosed, the facility will provide infrastructure for AI training, simulation, validation and deployment. The facility will have four floors covering 10,000 square meters to house several hundred robots.
The evolution toward autonomous data centers will progress through five stages: instrumented infrastructure, AI-assisted operations, autonomous physical inspection, AI-driven physical intervention, and ultimately self-optimizing facilities. At this final stage, integrated AI agents, robotics, digital twins, and autonomous controls will coordinate operations, anticipate failures, and continuously optimize performance within defined safety boundaries. Physical AI will be a key enabler of more resilient, efficient, and intelligent data centers.
Cloud providers embrace robotics
Cloud providers are increasingly exploring robotics to automate data center operations as infrastructure grows in scale and complexity. The focus is shifting beyond equipment transportation and inspection towards hardware maintenance, component replacement and lifecycle management. Meta and Microsoft are among the companies developing and testing robotic systems for these applications.
Meta has deployed autonomous robots for server rack transportation, inventory management and equipment inspections, while testing robotic arms for cabling, server power cycling and hardware handling.
Microsoft, through its Creative Robotics and Future Technologies (CRAFT) research programme, is developing robots capable of identifying and replacing optical transceivers, as well as performing optical-fibre cleaning and installation. Its longer term objective is to automate the hardware lifecycle, from rack assembly and deployment to maintenance, repair and decommissioning.
The development of data center robotics is also supported by wider advances in AI and industrial automation. Agile Robots SE and Google DeepMind have partnered to combine Gemini Robotics foundation models with industrial robotics hardware, targeting more adaptable robots capable of performing complex tasks.
NVIDIA and LG Group are similarly developing an AI factory to support the development, simulation and deployment of physical AI systems across robotics, autonomous driving and data centers. ABB Robotics and US start-up Molg are developing robotic microfactories to automate the disassembly and recycling of data center equipment, supporting component recovery and reducing electronic waste.
The data centers of the future will evolve into a highly coordinated physical AI ecosystem in which intelligent machines, autonomous robots, AI agents, and human engineers work together. Autonomous robots will increasingly support equipment inspection, security patrols, hardware logistics, and selected maintenance activities, while intelligent control systems optimize cooling, power distribution, and infrastructure performance.
More advanced applications may eventually include robotic server replacement, automated fiber-optic cable management, and humanoid robots performing selected maintenance tasks, although these capabilities are still emerging. The real opportunity is not simply replacing people with robots, but creating an intelligent, self-monitoring, self-optimizing, and increasingly self-maintaining data center.