Foreword:

MWC 2026 marks a watershed moment, as the event’s central narrative shifts decisively from [connection] to [intelligence].

Chinese companies are now leading the charge, transcending their traditional strengths in telecommunications equipment and consumer electronics to encompass the entire physical AI industry ecosystem—spanning computing infrastructure, network connectivity, core chips, precision components, industry solutions, smart terminals, and complete robotic systems.

The Invisible Foundation: Intelligent Infrastructure for the Physical AI Era

Unlike previous years, where robotics products were largely confined to laboratory demonstrations, this year’s MWC robots are designed with real-world industrial, commercial, and consumer applications in mind.

A profound industrial transformation is underway, with the global tech industry reconfiguring intelligent infrastructure to meet the systemic demands of physical AI.

Over the past decade, the mobile internet’s computing model has been highly centralized, with most processing occurring in remote data centers. User terminals were limited to information collection and display, while the cloud handled all complex computational tasks.

However, as AI begins to permeate the physical world, the limitations of this architecture have become fully apparent.

Robots navigating dynamic obstacles in complex environments require millisecond-level responsiveness. Industrial equipment demands deterministic, low-latency connections for real-time control. Vehicle-road coordination and remote operations rely on stable end-to-end communication.

These physical-world intelligence requirements simply cannot tolerate the delays and uncertainties introduced by cloud roundtrips.

Thus, a new computing architecture—featuring cloud-edge-device collaboration—has emerged as the cornerstone of the physical AI era.

The cloud handles large-scale training and global intelligent decision-making. Edge nodes perform real-time inference and data processing. Terminal devices are equipped with basic sensing and execution capabilities. Together, they form a complete computing system tailored for physical AI.

One of the key topics at MWC 2026 is how global companies are deploying computing power and networks around this architecture.

Huawei unveiled its next-generation AI computing solution, SuperPoD, overseas for the first time. Designed specifically to support the training and inference needs of embodied AI large models and multimodal physical-world models, it provides a robust computing foundation for physical AI.

Huawei also introduced the Agentic Core intelligent network architecture for the AI era, transforming traditional communication networks into intelligent operation platforms capable of supporting large-scale AI applications and enabling seamless connectivity between cloud “brains” and terminal “bodies.”

ZTE showcased a comprehensive AI-ICT integration system, launching its [AI Agentic Connectivity] framework, which deeply embeds AI capabilities into network architectures.

Through an agent-based approach, it achieves dynamic scheduling and self-optimization of network resources, enabling networks to autonomously adapt to the operational demands of physical AI.

Beyond this, innovations in underlying hardware—such as chips and modules—provide critical support for physical AI implementation.

ASR Microelectronics introduced a 6nm process intelligent SoC chip platform, delivering up to 20 TOPS of on-device AI computing power, enabling ordinary terminals to run complex AI models smoothly at the edge.

Longsys presented storage solutions for smart terminals, ranging from UFS 4.1 to the ePOP5x designed specifically for AI glasses, addressing the challenge of balancing space and performance in lightweight intelligent devices.

Module manufacturers like Quectel, Fibocom, and SIMCom launched intelligent module products integrating connectivity, computing, sensing, and control.

Some even extended downstream, offering complete solutions for humanoid robots and becoming key enablers for physical AI deployment.

From computing infrastructure to communication networks, from core chips to intelligent modules, Chinese companies have established a full-chain layout for physical AI infrastructure.

This means China’s physical AI development is not a serendipitous breakthrough in isolated technologies but a systemic advancement built on a complete and mature infrastructure ecosystem.

The Backbone: Full Industry Chain and Supply Chain Advantages

Behind a single humanoid robot lie high-precision servo motors, lightweight structural components, high-density batteries, multi-dimensional sensors, edge computing chips, and a manufacturing system capable of rapid iteration and assembly.

Only when these elements form a complete and efficient industrial chain can robots transition from laboratory prototypes to scalable industrial and consumer devices. This is precisely where China’s market holds its core strength.

Take Zhiyuan Robot as an example: over 70% of its core components are localized, with critical parts—from the drive modules of its dexterous hands to the structural components of its torso—supplied through the precision manufacturing ecosystems of the Yangtze River Delta and Pearl River Delta.

This highly mature industrial collaboration enables companies to maintain product performance while significantly reducing costs and accelerating iteration cycles.

Unitree Technology’s development path also exemplifies the unique advantages of China’s supply chain. Its G1 humanoid robot maintains high dynamic motion performance while keeping overall costs within a more market-accessible range.

Dubbed a “price dimensional strike” by the industry, this achievement stems from the cost-control and mass-production capabilities accumulated over years in China’s robotics supply chain.

While star robotics companies take center stage, numerous upstream supply chain enterprises are becoming the unsung heroes of China’s physical AI industry.

Lingyi Intelligent Manufacturing, a precision manufacturing giant that grew out of the consumer electronics sector, made its MWC debut by showcasing core robotics components such as harmonic reducers and planetary roller screws, along with dexterous hands and joint structural components manufactured using CNC and MIM processes.

As a key supplier to Zhiyuan Robot, BrainCo, and other enterprises, Lingyi Intelligent Manufacturing entered a strategic partnership with Zhiyuan in 2023 and began providing large-scale mass-production services in 2025.

Its Beijing Embodied Intelligence Mega-Factory has a planned annual capacity of 500,000 units and has already signed agreements with multiple robotics companies, including Magic Atom.

As physical AI moves from technical validation to large-scale deployment, cost control and delivery capabilities will increasingly become the core variables determining market success.

On this front, Chinese companies hold a relatively advantageous global position.

Over the past two decades, China has developed the world’s most mature and efficient precision manufacturing system and full industry chain collaboration capabilities in sectors like consumer electronics and new energy vehicles.

These capabilities can be seamlessly transferred to the physical AI domain, enabling Chinese companies to rapidly complete the entire process from R&D to mass production while continuously driving down costs through economies of scale.

AI is gaining a body—and increasingly, the bones, muscles, nerves, and blood for that body are coming from China.

The Physical AI Boom: An Inevitable Outcome of Development Paths

The European and American approach is technology-driven scenarios: achieve technical breakthroughs in labs, stack large model parameters, pursue algorithmic perfection, and then seek market applications for the technology.

This path did confer a leading edge during the generative AI era, as training data for generative AI was readily available online and did not rely on physical scenarios or infrastructure.

However, in the physical AI era, this path encounters insurmountable bottlenecks. Without ubiquitous networks, distributed computing power, or sufficient real-world scenarios, even the best technology cannot be deployed in the physical world.

China’s path, by contrast, involves infrastructure supporting scenarios, which in turn drive technological iteration.

Over two decades, China first built the world’s most complete new infrastructure, laying the “highways” of the digital world.

Subsequently, the digital transformation of thousands of industries generated vast real-world scenario demands, which then pushed continuous technological iteration and innovation.

This resembles the explosion of China’s e-commerce and express delivery industries a decade ago, which would not have been possible without the supporting transportation infrastructure of high-speed rail, highways, and rural road connectivity.

Without a nationwide road network, even the most advanced e-commerce models could not deliver goods to consumers.

Similarly, without Chinese-style new infrastructure, even the most advanced physical AI could not enter factories, mines, or the real physical world.

Conclusion:

The competition in physical AI is essentially a contest of system integration capabilities.

Whoever can integrate perception, computation, execution, connectivity, and energy management into a reliably functioning whole will win industrial dominance in the next decade.

Chinese companies’ comprehensive capabilities in this dimension are now being reevaluated by global peers.

Partial References: Industrial Insights: [Direct from MWC 2026: China’s Moment for Physical AI Has Arrived], Yicai: [Over 350 Chinese Enterprises Shine at MWC: From Smartphones to Robots, AI Goes Global Faster], Core Semiconductor: [Chips That Make Intelligence ‘Land’: This Year’s MWC Highlights], People’s Posts and Telecommunications News: [CES 2026 Reveals AI’s Next Destination: How ‘Physical AI’ Reshapes the Real World], Zhengjie Bureau: [From ‘Performance’ to ‘Business’: Chinese Robots Make a Collective Debut in Barcelona], QHOME: [In-Depth Analysis of Embodied Intelligence: MWC 2026 Marks the ‘Breakthrough Year’ for the Robotics Industry]