Unexpected AI: Physical Ai Device Shipments to Reach 145 Million Units by 2035

New and rapidly developing tech will bring artificial intelligence into real-world situations

The global physical AI market is entering a rapid growth phase as advances in robotics, edge computing, generative AI, vision technologies and sensor technologies increasingly enable machines to interact intelligently with the real world. Cumulative physical AI device shipments, including for vehicles, robots and drones, will reach 145 million units during 2025-2035, according to Counterpoint Research’s latest Global Physical AI Market Tracker report.

Talking about the AI market landscape, Counterpoint’s principal analyst Soumen Mandal said, “Physical AI represents the next major evolution of AI. While the first AI wave focused on digital intelligence – software that understands text, images and data, the next wave brings AI into the physical world, allowing machines to perceive their surroundings and interact autonomously.”

Counterpoint’s physical AI research covers multiple types of autonomous systems embodying spatial sensor-backed AI blended with a digital world. This includes self-driving vehicles, robots, drones and eventually newer form factors such as cameras. Within robotics, the service, industrial and humanoid segments will make up the bulk of the autonomous systems with embodied AI.

Service robots will account for the largest shipment volumes in the robotics segment, driven by expanding use cases across logistics, warehouses, hospitality, healthcare, cleaning, security and agriculture. Industrial robots, which currently have more limited deployment, largely concentrated in automotive, electronics and heavy machinery industries, where high system costs and complexity restrict volumes, will see wider adoption driven by broader applications, improving scale, lower costs and easier deployment models.

While still in the early stages of development, humanoid robots are gaining momentum as companies develop machines capable of performing complex human-like tasks across factories, warehouses and service environments. AGIBOT tops the global list of vendors with the highest number of annual humanoid robot installations, followed by Unitree, UBITECH, Leju and Tesla. The humanoid robot segment is expected to be the fastest-growing category in terms of shipments, with cumulative installations of humanoid robots projected to exceed 100,000 units by 2028, growing seven times over compared to 2025.

Commenting on the humanoid robot development, Counterpoint Research Vice President Neil Shah said, “Humanoid robots represent one of the most exciting long-term opportunities within physical AI.

Advances in generative AI, computer vision systems and motion control are bringing us closer to general-purpose robots that can operate in human environments. While there are advancements in the ‘form’, the ‘mind’ is something that is ripe for innovation. The industry has to cross the chasm from AMI (Autonomous Machine Intelligence) to embodied AGI (Artificial General Intelligence).”

Shah added, “We are closely monitoring the components, whether semiconductors, sensing or software, going into these robots. The rise of Vision-Language Models and Vision-Action Models unifies multimodal perception, language understanding and reasoning, and executable control within a single sequence modelling framework, which will be a critical inflection point. Although commercialization will take time, the long-term impact will be transformational across multiple industries.”

As physical AI systems gain more features and deeper real-world integration, mechanical component costs are likely to decline over time due to scale and maturity. However, the growing need for advanced computing capabilities will drive higher demand for compute and semiconductors, increasing their share of the overall system cost. The approach to this market varies across compute players.

NVIDIA is targeting the physical AI market with a data center-to-edge strategy, leveraging its strengths in AI training, simulation and high-performance compute platforms for robotics and autonomous machines. Qualcomm, meanwhile, is pursuing an ecosystem-first, power-efficient edge AI approach, focusing on integrated AI compute and connectivity platforms for robots, drones and other autonomous systems operating at the edge.

Commenting on opportunities for ecosystem players, Counterpoint Research Director Marc Einstein said, “Physical AI will create opportunities across the broader ecosystem. Beyond device makers, compute players will benefit by powering the ‘brains’ of these systems. Telecom operators will gain from increased data traffic, connectivity and edge services. Meanwhile, software and services providers will see recurring revenue opportunities through data analytics, lifecycle management, fleet services and cloud infrastructure.”

As physical AI systems scale across industries, the experts from Counterpoint Research assert that collaboration across the OEM, semiconductor, connectivity and software ecosystems will be critical to unlock their full potential. Companies that can build strong platforms and partnerships across the value chain will be best positioned to capture this emerging opportunity.