Cities that delay AI adoption risk falling behind permanently

A new study by Esteve Almirall of Esade Business School argues that the global race to harness artificial intelligence (AI) will not be decided by nations or corporations, but by cities. The research finds that urban ecosystems are becoming the primary sites where governance, mobility, and infrastructure are being fundamentally restructured.

The study, titled "Smart Cities in the Agentic AI Era: Three Vectors of Urban Transformation," published in Applied Sciences, explores how cities are entering a new phase of technological evolution driven by the convergence of agentic artificial intelligence, autonomous mobility, and urban robotics. The transformation underway is not incremental modernization but the emergence of a new urban order, comparable in scale to the Industrial Revolution, the study claims.

Three converging vectors are reshaping how cities function

The study identifies three distinct but interconnected technological vectors that are driving change across modern cities. These include the rise of agentic governance systems, the expansion of autonomous electric mobility, and the growing deployment of robotics within urban infrastructure.

• Agentic AI, defined in the paper as systems capable of reasoning, planning, and acting autonomously within structured workflows, is transforming public administration into what the study describes as cognitive government. These systems are increasingly able to process permits, manage services, and support decision-making processes in real time, shifting governance from static bureaucratic models to adaptive, data-driven systems.
• Autonomous mobility is redefining how people and goods move through cities. Robotaxis, autonomous buses, and delivery systems are not only changing transportation patterns but also acting as mobile data platforms, continuously collecting high-resolution information about traffic, infrastructure, and environmental conditions.
• Urban robotics extends automation into the physical fabric of cities. From maintenance robots to inspection drones, these systems are increasingly integrated into infrastructure management. Their value, the study notes, is not in isolated deployment but in their ability to operate as interconnected nodes within a larger intelligent system.

What makes these vectors transformative is not their individual capabilities but their interaction. When governance systems, mobility networks, and robotic infrastructure operate in isolation, their impact remains limited. When they converge within the same urban ecosystem, they begin to reinforce each other, creating entirely new forms of functionality and efficiency.

Cumulative recursive hybridisation creates compounding urban advantage

The study discusses the concept of cumulative recursive hybridisation, a framework that explains how cities generate accelerating returns when these three vectors are deployed together. Based on systems thinking and historical parallels with industrialisation, the research argues that urban transformation occurs through iterative cycles of interaction across multiple domains. Each technological advance feeds into others, creating feedback loops that compound over time.

The study identifies four key loops that drive this process: data, regulation, infrastructure, and talent.

• Data loop: It is powered by the continuous collection of information from autonomous systems. Vehicles, robots, and sensors generate streams of real-time data that improve predictive models and operational decisions. As systems become more effective, they generate even more data, reinforcing the cycle.
• The regulatory loop reflects the ability of city administrations to adapt rules in real time. Cities with agile regulatory systems can accelerate the deployment of new technologies, enabling faster experimentation and iteration.
• Infrastructure loop: Here, intelligent systems enhance the physical environment, which in turn supports further technological deployment. Robotic maintenance and smart infrastructure increase efficiency and reliability, creating a foundation for continued innovation.
• The talent loop highlights the role of human capital. Cities that attract engineers, developers, and entrepreneurs create dense ecosystems where knowledge circulates rapidly. This concentration enables cross-fertilisation across domains, leading to new hybrid solutions that would not emerge in isolated environments.

Together, these loops produce a path-dependent dynamic. Cities that activate all four simultaneously gain compounding advantages, while those that delay adoption face increasing difficulty catching up. The study draws a direct parallel with the Industrial Revolution, where technological breakthroughs did not occur in isolation but emerged from dense urban clusters where innovations interacted and reinforced each other.

Cities, not nations, are the true battleground of AI transformation

City, rather than the nation-state or corporation, is the natural unit of analysis for understanding AI-driven transformation. This is because cities are where regulatory authority, infrastructure management, service delivery, and innovation ecosystems intersect. At the municipal level, decisions about licensing, data governance, and infrastructure investment directly shape how technologies are deployed and integrated.

The research argues that institutions must reorganize themselves to reflect the capabilities of new technologies. In the case of agentic AI, this means cities must adapt their governance structures to accommodate not only human actors but also AI agents with defined roles and decision-making authority.

Cities that successfully align their institutional structures with technological possibilities can unlock significant value. Those that fail to do so risk falling into what the study describes as an asymmetry trap, where outdated governance models prevent them from fully leveraging new capabilities.

Failures reveal the limits of smart city ambitions

While the study highlights the potential of agentic AI-driven urban transformation, it also examines cases where such efforts have stalled or failed, underscoring the importance of institutional and social factors.

The collapse of the Sidewalk Labs project in Toronto demonstrates how public resistance and governance challenges can disrupt even well-funded smart city initiatives. In this case, the inability to secure trust and regulatory alignment prevented the necessary feedback loops from forming.

Similarly, the rollback of autonomous vehicle operations in San Francisco shows that regulatory and safety concerns can reverse progress, highlighting the bidirectional nature of feedback loops. Gains are not guaranteed; they depend on maintaining public acceptance and operational reliability.

The study also points to the limitations of master-planned cities such as Songdo, where advanced infrastructure failed to generate the expected innovation ecosystem due to a lack of organic talent density and knowledge circulation.

These cases clearly suggest that technology alone is insufficient. Successful transformation requires the simultaneous alignment of governance, infrastructure, talent, and social legitimacy.