Human-Centered Physical AI: Reconstructing Future City Operating Systems

When Urban Systems "Break the Chain" During Morning Rush Hour

At 7:42 a.m., on a main artery of a large city, a burst water main closes a key intersection. Buses are unable to get real-time traffic conditions and have to make temporary detours; logistics trucks queue up in congestion at the port exit; navigation apps crash due to a sudden surge in traffic. Tens of thousands of commuters are stuck on the road—not because the city lacks infrastructure, but because systems are not connected and lack resilience.

This is the scenario depicted in a recent World Economic Forum article on physical AI. It reveals a fundamental problem: cities are complex physical systems, but the current level of digitalization is far from sufficient to support their efficient operation. The World Bank points out that more than 80% of global GDP is generated in cities, and the United Nations projects that by 2050, 68% of the world's population will live in cities. The mismatch between growth and planning is threatening the sustainable development of cities.

Physical AI: Equipping Cities with a "Real-Time Nervous System"

Traditional AI is mostly applied in the digital world, such as optimizing IT budgets or recommending content. But cities need real-time management of the physical world—roads, pipe networks, buildings, and public spaces, which are changing every minute. The concept of physical AI was born from this.

Physical AI combines the Internet of Things (IoT) with artificial intelligence, but the key difference lies in architecture: computation happens at the edge, not in the cloud. By processing data locally, cities can achieve faster response times and lower latency. These AI models are lightweight, cost-effective, and based on physical and mathematical principles, rather than language models. This enables physical AI to manage infrastructure continuously and in real time, delivering a "step change" in responsiveness, reliability, and service quality.

From air quality monitoring to traffic flow regulation, from public safety warnings to fire prevention, physical AI covers every aspect of urban operations. As autonomous vehicles gradually enter urban environments, reliance on physical AI will grow exponentially. Madrid's Chamartín railway station is a typical case, showing how physical AI can be used to optimize the flow of people and goods.

Breaking Down Silos: Innovation Ecosystems and Financing Mechanisms

But technology alone is not enough. Urban transportation, energy, logistics, and public service systems have long been managed separately by traditional institutions, creating data silos. For physical AI to work, an integrated innovation ecosystem must be established—one that brings together businesses, academic researchers, and municipal leaders to co-create solutions. Such collaboration must not only generate social value but also deliver economic returns, making cities engines of efficiency, sustainability, and inclusiveness.

At the same time, AI is expensive, as the latest valuations and the data center boom have demonstrated. Cities need new financing mechanisms to modernize aging systems and pilot emerging technologies. The combination of public and private capital can reduce risk and expand scale. For example, the "infrastructure as a service" model could become a way for cities to acquire AI capabilities.

Digital Connectivity: The "New Infrastructure" for 21st-Century CitiesJust as 20th-century cities relied on roads and electricity, 21st-century cities depend on connectivity. Broadband, 5G, IoT networks, and AI platforms form the infrastructure for future urban operations. Without these, smart transportation, digital energy networks, and real-time public safety would be impossible.

Many innovative cities have already demonstrated the possibilities. For example, through physical AI as a service, reimagining the way goods move between ports, rail hubs, and distribution centers. With collaborative planning, data-driven routing, and smarter rail connections, cities can significantly reduce congestion and emissions in freight transport.

People-Centered: Technology Must Serve Human Needs

Ultimately, AI itself will not create great cities—people will. All innovations—including digital twins and physical AI—must serve human needs such as accessibility, safety, equity, sustainability, and quality of life. This is the essence of "people-centered urban design": leveraging technology and financing strategies to improve the experience of every citizen, regardless of where they live or how they travel.

But a growing challenge has emerged in reality: the rapid expansion of data centers in densely populated areas around the world is putting enormous pressure on power grids that were not originally designed with sufficient capacity, leading to grid congestion and rising electricity prices for residents. This raises deeper questions about the "power equation of the AI era."

Put simply, AI capability is positively correlated with access to electricity: the less energy available, the more constrained innovation becomes; the greater the capacity, the more pronounced the acceleration. Without carefully designed regulation, this could create urban inequality—big tech companies gaining priority access to energy while communities bear higher costs and receive limited local benefits.

Redefining the Relationship Between Cities and Computing Power

Cities are not without options. One possible solution is: for all data centers located within city boundaries, reserve a "local offload" channel for edge-based physical AI applications. These data centers would benefit by delivering immediate value to local communities. This both alleviates grid pressure and keeps the technology dividends within the city.

Cities are catalysts for opportunity, innovation, and community. By combining creative financing with integrated innovation ecosystems, cities can not only reimagine their own economies and mobility but also achieve a more connected, inclusive, and resilient future. This is the vision that makes physical AI truly worth anticipating—it is not merely a technology upgrade, but another evolution of urban civilization.