As highlighted in global smart city roadmaps and institutional studies from the World Economic Forum and the OECD Roundtable on Smart Cities and Inclusive Growth, sustainable AI for cities can play a transformative role in building resilience, mitigating climate change effects, and supporting sustainability targets. Examples include adaptive traffic signal systems that remarkably reduce CO2 emissions and congestion, and AI-powered infrastructure monitoring that saves resources and prevents failures before they occur.
Pilot programs in cities worldwide back these claims with data. For instance, Barcelona’s smart city initiatives have achieved significant cost savings through IoT systems —$58 million in water management and $50 million annually in parking revenues according to Harvard’s Data-Smart City Solutions. In Singapore, AI monitoring and optimization systems for energy distribution of buildings led to 15% energy savings. These are not isolated successes, but proof of AI’s potential to support meaningful environmental and economic change.
Despite these achievements, a critical question remains: are sustainability goals truly driving AI innovation, or are they still seen mainly as risk-mitigation measures rather than growth opportunities?
The dual lens of sustainable AI
To understand what sustainable AI for cities truly means, we must first acknowledge its dual nature. It builds on ethical principles and encompasses the environmental, social and economic dimensions of sustainability, ensuring AI benefits not only in the short-term, but also in the medium and long-term.
As a result, AI sustainability does not consist just in developing new technologies for sustainability needs, as frequently perceived. It comprises two complementary and essential dimensions:
- Developing AI solutions that address sustainability challenges (e.g., reducing emissions, optimizing energy and water use);
- Ensuring that AI systems themselves are sustainable — ethically, environmentally, socially, and economically.
Table 1. AI key impact to accelerate the climate transition
Source: What is AI’s role in the climate transition and how can it drive growth?
The dual approach is essential to leverage AI full potential while mitigating emerging AI risks and ensuring its efficiency and scalability beyond the short-term.
Indeed, studies have recently highlighted growing environmental costs of AI systems and supporting infrastructures, including rising GHG emissions, water and material consumption, and e-waste. For instance, Google’s GHG emissions rose 48% from 2019 to 2024 due to AI training and data center expansion.
Source: Power Hungry Processing: Watts Driving the Cost of AI Deployment?
Data center expansion for AI has also significant environmental implications for cities, as data centers are increasingly located near urban areas to provide low-latency services and high-performance computing. This adds to other emerging social and business concerns such as AI manipulation, security and privacy issues, limited human control, discrimination and amplification of existing stereotypes, and unequal business conditions.
Table 2. Mean and standard deviation of energy per 1,000 queries for the ten tasks.
Source: Power Hungry Processing: Watts Driving the Cost of AI Deployment?
Beyond risk: enabling innovation through sustainable design
Despite common perception, AI sustainability does not adopt a defensive stance that limits AI research and innovation. Instead, it funnels AI innovation efforts in ways that maximize overall AI benefits for users, stakeholders, businesses, and the environment by assessing positive and negative AI system impacts. This approach provides IT and business professionals with valuable insights to unveil new opportunities for system innovation and to enhance service’s competitive advantage, while building a safe impact-driven AI system.
Viewing AI sustainability solely as a risk-mitigation exercise—often narrowly focused on energy consumption—overlooks the broader environmental, social, and economic impacts and interlinkages across the entire AI system lifecycle. This limited perspective can reduce opportunities for innovation, while increasing exposure to misuse, unintended or harmful outcomes, and security breaches. At the same time, the competitive AI market places intense pressure on businesses to rapidly reach the market and gain a competitive edge, often limiting comprehensive risk assessment and testing. This increases chances of drawbacks, occasionally forcing system withdrawals (e.g., AWS AI recruitment tool withdrawn after exhibiting gender bias).
It is worth noting that in today’s short-term market logic, AI solutions for climate mitigation and urban sustainability—despite their high potential shown by city pilots—frequently result less economically attractive than applications with quicker return of investments, due to their higher technical and business complexity, reliance on larger investments and long-term public–private collaborations. It is essential to support AI innovations for city sustainability and climate mitigation through business and financial incentives and training.
Aligning AI strategy with societal and political priorities
Global AI competition is influencing not only businesses but also governments to make strategic investments aimed at securing technological leadership. For example, the EU announced in April the AI Continent Action Plan, which outlines a comprehensive strategy to position Europe at the forefront of AI innovation through coordinated investments in infrastructure, data ecosystems, algorithm development, skills, and regulatory simplifications. While government AI investments are both necessary and vital, they should prioritize strategic societal and environmental needs as our resources and time are limited.
Furthermore, current geopolitical instabilities and economic uncertainties are diverting AI efforts and investments away from sustainability targets—not only due to the intensification of AI innovation in defense and military applications, but also because of the growing focus on dual-use AI research, which supports both civilian and military purposes. It is worth noting that steering early-stage AI research toward military objectives raises ethical concerns and questions about its alignment with sustainable development.
Conclusion: a sustainable AI ecosystem for cities
To ensure a positive long-term impact, we must adopt a systemic, sustainable approach to AI — one that integrates environmental, social, and economic perspectives, embraces stakeholder collaboration, and supports flexible governance and ethical design from the start.
Sustainable AI for cities is not an abstract concept. It requires clear priorities, incentives, transparent governance, and widespread engagement from business leaders, urban planners, policymakers, educators, and the media. Universities and research hubs play a key role in building responsible AI ecosystems through transparency, accountability, and long-term thinking.
Sustainability must be more than a framework — it must be the compass guiding how we develop and apply AI in an increasingly complex and fragile world.
About the author: Daniela Tulone, PhD, Founder and President of ecoSurge, EU expert
Read more about smart city and urban intelligence here, here and here.
