Smart cities failed. Sentient cities are coming next

John Rossant is CEO and Founder of CoMotion, a global platform connecting the architects shaping the future of mobility and cities. CoMotion GLOBAL convenes in Riyadh, Saudi Arabia, on December 7-9.

In 2015, Columbus, Ohio won the U.S. Department of Transportation's $40 million Smart City Challenge. By 2020, the promised transformation had not materialised. Traffic remained congested. Systems didn't integrate. Citizens saw incremental improvements, at best.

Columbus isn't an outlier. It's the norm.

For a decade, the smart city promise has tantalised urban planners. We imagined interconnected, efficient metropolises managed by data streams and dashboards. Instead, we got sophisticated gadgets that generate reports but struggle to prevent problems.

Kansas City spent millions on smart streetlights that collect data. Barcelona's celebrated infrastructure raised surveillance concerns without proportional livability gains.

The fundamental flaw? These systems are reactive, not anticipatory. A sensor detects a water main break after thousands of gallons are lost. A traffic monitor reports congestion when commuters are already stuck.

The infrastructure required for the 21st century – defined by climate volatility, rapid population shifts, and autonomous systems – demands more than post-facto reporting. It demands prediction and adaptation.

Technical leap: From sensors to prediction

What's emerging is infrastructure powered by Large World Models and geospatial AI – systems that understand context, predict outcomes, and optimise across multiple variables simultaneously. We call these sentient cities.

Where traditional infrastructure reacts to a sensor reading, AI-native systems identify cascading traffic failures hours before they occur – noticing a sporting event will end during rush hour, construction is funneling vehicles toward a bottleneck, and weather will slow speeds.

The system doesn't just report this; it adjusts signals across the metro, pushes transit alerts, and coordinates parking systems.

Singapore's mobility systems already use machine learning to optimise traffic lights in real-time, reducing travel times by 15% during peak hours.

Copenhagen's heating system uses AI to predict demand 48 hours ahead, cutting energy waste by 25%.

Testbed opportunity

Most cities can't easily deploy next-generation infrastructure. Political fragmentation means transit, utilities, and safety operate in incompatible silos. Union contracts and risk-averse bureaucracies slow innovation.

This creates opportunity for cities to make bold bets.

For example, under Saudi Vision 2030, Riyadh is building new infrastructure with AI integration as a core design principle – leapfrogging the slower, incremental upgrades typical in older, denser cities.

The Kingdom is channeling substantial investments into sovereign AI infrastructure, advanced computing, and digital twins, enabled by strategic partnerships with global technology leaders.

Dubai has committed over $150 billion to AI-powered infrastructure by 2031. Songdo, South Korea, continues evolving its integrated systems.

These testbeds can generate insights transferable to constrained, legacy environments – the reality for most cities.

The governance challenge

But there are also hard questions to answer. If infrastructure can predict and respond autonomously, who sets priorities?

When algorithms optimise traffic, do they minimise aggregate travel time or ensure equitable access for distant low-income neighbourhoods? When predictive systems identify "high-risk" areas, how do we prevent feedback loops that over-police communities of colour?

The risks aren't hypothetical.

ProPublica found risk assessment algorithms twice as likely to falsely flag Black defendants as high-risk.

Sidewalk Labs' Toronto project collapsed partly due to data governance concerns. Well-intentioned algorithms can encode biases in training data, perpetuating inequities at software speed.

Cities must own their data and models. Contracts must ensure cities can audit algorithms and maintain control if partnerships end.

There must also be algorithmic transparency: Black-box AI is incompatible with democratic governance.

And communities most affected by infrastructure decisions must be involved in designing systems, not just feed back after deployment.

This December, we're convening the Sentient Cities Task Force at CoMotion GLOBAL in Riyadh, bringing together planners, AI researchers, privacy advocates, and officials to develop shared frameworks around data governance and ethical AI deployment at urban scale.

By connecting the ambitious work happening in the Middle East with global expertise, we can ensure the next generation of urban infrastructure is built with equity and transparency as core principles, not afterthoughts.

A path for legacy cities

For cities anchored by century-old infrastructure and tight budgets, the answer isn't wholesale replacement. It's strategic augmentation.

Start with maintenance. Predictive systems analysing sensor data from water mains and bridges forecast failures before they happen. Pittsburgh reduced water main breaks using predictive analytics.

We must also optimise existing systems: Los Angeles synchronised traffic lights using adaptive algorithms, reducing travel times 13% without adding road lanes.

We should also embrace private-sector innovation while maintaining public ownership of data. Barcelona's model prioritises data sovereignty and open-source platforms.

The transition won't be easy or cheap. But the cost of inaction is rising. Climate change strains infrastructure designed for last century's weather and ageing systems fail more frequently.

The gap between cities with predictive, AI-driven infrastructure and those relying on reactive legacy systems will widen rapidly and will manifest in quality of life, economic competitiveness, climate resilience, and fiscal sustainability.

The era of the smart city is ending. What comes next is infrastructure that anticipates, adapts, and optimises in real-time.

The question is whether cities will embrace this shift thoughtfully – with attention to equity, transparency, and democratic governance – or recklessly, creating systems that optimise efficiency while eroding trust and deepening inequality.

The cities that get this right won't just be smarter. They'll be more livable, sustainable, and resilient.