Issue

AI 02 Cities 02: Energy

for Future

AI for Future Cities

Energy
Arup
May 2025
1
AI for Future Cities Series

The rapid evolution of artificial intelligence (AI) promises to enhance

efficiency, speed and innovation. But what does this mean for cities:
for how they are planned, designed, built and managed? What are
the opportunities and what are the risks? Amongst the hype and
speculation, how do we ensure that these technologies support us in
achieving our shared goal of creating cities that are better for people
and the planet?
Compiled by Arup’s Foresight team, AI for Future Cities is a series
of critical reflections and expert insights on the uses and impacts of
artificial intelligence across all aspects of our cities – from planning
through to running them, from infrastructure through to the flows of
resources. Rather than an in-depth and comprehensive report, the issues This is issue 02 of a series
are designed to be read as expertise-led explorations and provocations. on AI for Future Cities.
Upcoming publications will
It will give you a rich understanding of how AI already operates in the focus on Water, Circularity,
urban context today, what trends are shaping its use tomorrow, and Nature and Workplace.
informed speculation on the long-term possibilities. Explore other issues here.
AI for Future Cities 02: Energy

Contents
This second issue in the series focuses on energy: How is AI changing
the way that energy is consumed and produced in our cities? How
might new, AI-enabled energy systems change the physical fabric
of our cities and the lived experience of their residents? What new
competencies may future designers require for creating and working
with energy systems that have been augmented by AI? And how do we
ensure that AI remains a tool we use intelligently?

Foresight perspective 05

Horizon scan evidence: Trends shaping the future of AI in urban energy 07

The city of 2035: How might AI shape the city, and what new realities emerge? 08

Expert insight: Interoperability is required to unlock AI’s full potential 09

Case studies: What are we doing today to make cities better through AI? 11

Expert insight: AI is key to the transition to carbon-free energy 12

Case studies: What are we doing today to make cities better through AI? 14

Recommendations 15

References and Contributors 18

3
AI for Future Cities 02: Energy

Foreword
Meeting our energy challenges with AI
Nearly three years have passed since the explosion
of excitement around AI, sparked by the emergence
learning to reduce energy consumption in buildings at
scale. Our work on the UK’s Virtual Energy System
We face huge sustainability and
of ChatGPT. Since then, surging investment has Programme and datasharing infrastructure is laying resilience challenges alongside
triggered a proliferation of different AI models that the foundations to implement AI into our grids and, in
can improve how we work and what we deliver. so doing, helping to incorporate renewable energy and the ever-pressing need to deliver
the large amounts of storage they require to balance infrastructure and prosperity at
Applying AI to the real world is the the grid. In transport, AI enables the vast quantities of
opportunity batteries in our electric vehicles and bus fleets to be pace for communities around the
used as back-up energy storage for our urban energy
This series by Arup University’s Foresight team
explicitly focuses on applications of AI that systems, while supporting better transport demand world. It’s difficult to imagine
have tangible, real-world benefits in accelerating management, passenger flows and transport network how we can meet these demands
decarbonisation and supporting climate resilience. operation and maintenance.
This issue focuses on urban energy, a topic at the crux successfully without the power of
of two trends shaping our future world, electrification
and urbanisation. AI can play a significant role in new
AI to enhance human innovation,
ap-proaches to energy generation and distribution creativity, and productivity.
in cities, and we are already using it to plan, site,
and operate energy infrastructure. We’ve worked
on feasibility studies that prepare the UK’s grid
for the integration of AI, and we’ve used machine

Will Cavendish
Global Digital Services Leader
Arup

4
AI for Future Cities 02: Energy

Foresight Perspective:
How can AI help us manage the rapidly changing shape of
the grid and the needs of the city?
Modern cities are unimaginable without electricity. Now, this is changing with the energy transition to net AI is already transforming urban energy
Abundant and affordable energy has enabled us zero well underway in some areas of the world, and AI is emerging as the thread connecting these
to build upwards and outwards. Electric lifts have just beginning in others. new energy assets and needs at multiple scales,
shaped our skylines; electrified mass transit systems transforming energy into a flexible, dynamic layer
allow millions (if not billions) of people to live Technological advances in renewables mean that of the urban fabric rather than rigid infrastructure.
miles away from where they work and play, and energy can be generated cleanly, in a decentralised At the national scale, AI can help to plan where best
abundant lighting has enhanced urban nightlife and way, at competitive prices; and it’s becoming to install energy generation and forecast with greater
safety. commercially viable to store distributed electrical accuracy how much electricity can be produced on
energy thanks to innovations in battery technology. any given day, helping to balance the grid overall.
Yet, while energy systems are fundamental to our This means that achieving a net-zero grid and At the urban scale, AI turns cities and residents
lives and cities, the role of the city in the energy decarbonising our cities’ energy is more and into active nodes in the electricity grid rather than
system has barely changed in a century: electricity more a social and political challenge rather than a destinations. The management of fluctuations through
is generated at thermal power stations and cities technological one. Progress will depend on our ability distributed, dynamic storage and generation in homes
consume it. As a result, cities consume three and willingness to integrate clean, variable, spatially- and workplaces is made manageable with AI: it helps
quarters of global primary energy supplies.1 They distributed energy generation as well as incorporating to integrate buildings’ BESS (battery energy storage
remain the main final node of an energy system that whole new areas of decentralised energy demand systems) and EV (electrical vehicle) batteries into the
has been historically centralised and predictable. (from transport to heat to industrial electrification), to grid, control rooftop solar panels, optimise heating
new forms of connected storage in private homes and and cooling networks.
cars — all the while balancing the grid and keeping it
resilient in the context of a changing climate. 1. IEA (2024), Empowering Urban Energy Transitions, IEA, Paris https://www.iea.org/
reports/empowering-urban-energy-transitions, Licence: CC BY 4.0

5
AI for Future Cities 02: Energy

Reimagining urban energy through AI demand. These are just some of the possible future
opportunities that AI in energy unlocks for our cities.
“ Imagine a city where AI is
AI is the key technology that is driving the urgent
transition from managing a century-old system of
fully integrated into the energy
Accelerate, but beware of risks
centralised and predictable energy production towards
But the integration of AI in the urban energy system system, where buildings and
adaptation for decentralisation, where the city and its
residents are both consumers and nodes.
comes with new risks. AI is itself a drain on power neighbourhoods generate and
grids, and we must ensure that its deployment is
net positive. Similarly, relying on AI for decision- exchange their own power, and
Imagine a city where AI is fully integrated
into the energy system, where buildings and
making in the operation of critical infrastructure
might introduce new cybersecurity vulnerabilities.
it happens autonomously in the
neighbourhoods generate and exchange their own
power autonomously as life ebbs and flows around
Additionally, like any technology, AI can be background.”
misappropriated by actors who may use it to optimise
them. In this city, surplus revenue from energy for outcomes that run contrary to the benefit of the
can subsidise poorer residents or be invested in wider system and public.
community initiatives. Imagine a city where AI-
enabled energy decentralisation makes consumers
How can we ensure the right data sharing and
more resilient and where grid operators, drawing on
interoperability between assets and actors exists to
cross-sector forecasts, have complete transparency
enable AI’s full potential in our urban energy systems?
over energy supply and demand to anticipate near
How can we use AI to make the most of investments
and far-future needs. In this city, we will be able to
into EVs, heat pumps, solar and home storage? What
contain disruptions caused by natural disasters, to
new skills and responsibilities do stakeholders in our
blackouts or to cyber-attacks; backup energy supplies
energy system need to prepare for today?
could be brought online more rapidly. And imagine
the city grid using AI to interface with smart buildings
and infrastructure to predict and direct supply and
6
AI for Future Cities 02: Energy

Horizon Scan Evidence:

Trends shaping the future of AI in Energy

This is just a small selection 02 06 23 30 31

of key data we are tracking on

how AI operates in the context
of cities at present (now→),
of emerging trends we are
observing (near →), and of
informed speculations we
are making about long-term
possibilities that stem from
signals of change (next →).

02 03
01
04

08 16 26 39 42

Get in touch to explore our

expanding database of trends,
spanning across all issues of
the AI for Future Cities series.
7
AI for Future Cities 02: Energy

The City of 2035:

How might AI shape the city, and what new realities emerge?

8
AI for Future Cities 02: Energy

Expert Insight:
Interoperability is vital to unlock AI’s full potential
The emerging convergence of unpredictable demand, In this complex landscape, AI with its real-time of urban systems, facilitating optimised resource
scarce resources, and the need for sustainable data analysis and predictive capabilities is no longer allocation and adaptive infrastructure development.
development is driving a significant reimagining of a futuristic novelty but a vital tool for addressing
urban and industrial planning, including their energy current and future challenges. Intelligent grid management systems, powered by
needs. machine learning, can forecast demand, optimise
Specifically in the energy sector, transformation is energy distribution, and efficiently integrate
Population fluctuations, economic underway. The shift towards decentralised energy
generation, driven by renewables like solar and wind,
renewable sources. This is not merely aspirational but
a present technical and social necessity. Without AI,
and lifestyle shifts, and climate introduces supply variability. Simultaneously, demand maintaining grid stability and ensuring reliable energy
change present challenges that patterns are becoming unpredictable, influenced
by factors such as electric vehicle adoption and
supply amid these challenges becomes exceedingly
difficult.
traditional planning models were contemporary lifestyle patterns.
not designed to address. AI is essential for balancing this dynamic equation in The benefits of interoperable
energy and other areas of urban planning. models extend beyond grid
efficiency.
Unlocking AI’s full potential
The true potential of AI depends on interoperability.
When AI systems operate in isolation, confined to
specific sectors or projects, they create and operate
in silos of data, hindering comprehensive planning.
Lyndon Ruff
AI Centre of Excellence Manager Collaborative frameworks that enable industries to
National Energy System Operator share data and develop compatible AI platforms are
Magnus Cormack crucial for enabling a dynamic and responsive system.
Senior Manager, AI & Analytics This interoperability fosters a holistic understanding
National Energy System Operator
9
AI for Future Cities 02: Energy

When different sectors seamlessly exchange data In conclusion, the challenges facing the built
and insights, the potential for innovation and environment, energy sector, and urban planning
efficiency multiplies exponentially: supply chains demand a shift towards AI-driven solutions. However,
can be optimised, resource consumption minimised, realising AI’s full potential hinges on fostering
and circular economy principles more effectively interoperability and compatibility across sectors and
implemented. For instance, integrating AI-powered systems. This means greater collaborative efforts
logistics with smart manufacturing systems can and a commitment to data sharing that can enable
reduce waste and boost productivity. Similarly, energy resilient, sustainable, and equitable urban
information exchange between the building industry environments for the future.
and energy sector can lead to more efficient future
planning, design and operation.

The need for shared standards

Achieving this level of interoperability requires
concerted efforts from industries, governments,
and research institutions, but establishing open and
shared standards and protocols for data sharing, such
as NESO’s Data Portal and the energy Data Sharing
Infrastructure, is key. Alongside interoperability of
data, AI model compatibility is essential. Ultimately,
the goal is to achieve connected digital twins
representing the system as a whole, as envisaged by
our ‘Virtual Energy System Programme’, or the UK’s
National Digital Twin Programme.

10
AI for Future Cities 02: Energy

Case Studies:
What is Arup currently doing to make cities better through AI?
Arup has worked with partners and
Digital Spine Feasibility Study
clients around the world on AI-driven United Kingdom, 2023
Read more

solutions for real-world application that Arup collaborated with Energy Systems Catapult and the vital component in the decarbonisation of Great Britain’s energy
system. Ultimately, this will reduce the barriers and friction
deliver better outcomes for cities. University of Bath to explore the concept of a ‘digital spine’
of data sharing. With better, assured data-sharing and cross-
identifying the needs and challenges for the energy sector to
facilitate data sharing through a digital infrastructure. network collaboration enabling a more efficient, responsive flow
These cutting-edge projects give us The report, commissioned by the UK Government’s Department
of energy around the network and more intelligent balancing of
the system, a host of benefits will be unlocked for consumers,
powerful insight into how AI may for Energy, Security and Net Zero (DESNZ) came as a result of businesses, and energy companies.
the Energy Digitalisation Taskforce and its recommendations to
shape the future expectations of develop a ‘digital spine’ for the sector. The study highlighted the The concept of data sharing infrastructure detailed in the study
practitioners and how these may be challenges for the energy sector in facilitating data sharing and now serves as the foundation for Arup and NESO’s Virtual
how these challenges could be overcome through an enabling Energy System Programme, which seeks to implement the data
transferable to other domains across the infrastructure. This energy sector data sharing infrastructure will sharing infrastructure in the real world.
city. How might open data and data infrastructure provide a secure and resilient method of sharing data, which is a
introduce new opportunities for efficiency,
sustainability and innovation?

Virtual Energy System Programme

Read more
United Kingdom, 2022 – present

Arup is supporting Britain’s National Energy System Operator processes, data and technology that will be crucial to realise
(NESO) in developing the Virtual Energy System Programme, NESO’s Virtual Energy System. The result is a pathway to
the world’s first ecosystem of connected digital twins for a collaboratively develop the governance, data standards, access
national energy system. This will enable data-driven decisions policies, security, skills, and wider support required to efficiently
and whole-system monitoring, as well as supporting operational deliver the range of use cases that comprise a working Virtual
optimisation, while ensuring an efficient and reliable energy Energy System.
system fit for the future. It was also clear from this work that the day-to-day operational
Our work began at the organisational level, focusing on a insights and future planning that the system depends on –
framework that could power this large-scale transformation. such as the location and change of supply and demand – will
Through research, expert interviews, and extensive industry- increasingly rely on a common, industry-wide Data Sharing
How might AI-enabled digital twins change the way wide engagements, we developed a ‘common framework’: Infrastructure.
we model and optimise energy grids? we identified 14 key socio-technical factors, across people,

11
AI for Future Cities 02: Energy

Expert Insight:
AI is key to the transition to carbon-free energy
AI is both a challenge and an opportunity for the The energy industry is at a point of inflection For operators, use cases have included identifying
energy industry. On the one hand, the boom in data unprecedented in recent history. The rapid demand signatures of appliances on the grid to
centres to power AI models is leading to electricity decarbonisation of everything means electrification inform projections of electrification and using data to
load growth for the first time in years in the United of vehicles and heating systems is occurring proactively provide warning of potential transformer
States, Europe and other industrialised nations. This simultaneously with a boom in energy consumption or grid equipment failure. We are now seeing these
rapid pace of growth is challenging the traditional for data centres. This incredible confluence requires use cases expand to include computer vision to
grid structure and the speed at which new generation multiple solutions and there is no silver bullet. predict the risk of extreme events, such as wildfires
can be brought online. On the other hand, AI has that might damage grid infrastructure from remote
the potential to completely revolutionise planning sensing and drone imagery. This use of AI makes
and operation of the electricity grid, enabling However, with more solutions the grid safer and more reliable, with the potential to
truly right-time interactive models of distributed comes more complexity, and our prevent tragedies such as the Camp Fire in California
energy production and consumption that respect in 2018.
the complexity of cities and the needs for energy of traditional one-way power flow
buildings, vehicles, and more in a sustainable future. models may no longer be effective At Arup, we have used semantic interoperability and
or efficient. machine learning to help enable system flexibility
and predict the most cost-effective ways of reducing
energy consumption for buildings and cities even
Enter AI as they begin to electrify. In the US, our work with
AI is already rapidly elevating how we model and Whole Foods and National Resources Defense
optimise energy systems to take advantage of greater Council (NRDC) used machine learning to train
interactivity and more complex and intermittent models to search for novel combinations of energy
Rob Best sources of demand and supply. For the past 20 years, conservation strategies to achieve deep reductions
Americas Digital Energy Leader we have used machine learning models to both help in building energy consumption within cost and
Arup
plan demand and proactively identify grid conditions deployment constraints. This work showed that
Simon Evans that could lead to failure. machine intelligence coupled with human review
Global Digital Energy Leader
Arup

12
AI for Future Cities 02: Energy

and guidance could lead to up to 10% greater savings historic data and identifying novel relationships of solar and wind. This truly interactive and digitally
per building within the same budget. While this between behaviour, external conditions, and the grid interoperable energy system is within reach if we
may sound small, an additional 10% savings per response, AI can help grid operators identify and blend the creativity of human engineers with the
building means far less new generation required on deploy distributed resources at a scale that alleviates expansive potential of data and AI.
the grid to meet surging electricity demands. In the local distribution and transmission investment.
UK, we have developed a semantic data model for
energy flexibility markets that provides transparency Zooming out, grids are experiencing significant delays
and standardisation, supporting information layer in interconnecting new renewable generation. A core “AI has the potential to rapidly
interoperability that is required for scalable data driver of these delays is the analysis and testing of
exchange and effective use of AI. a variety of failure conditions, but AI solutions are
elevate how we model and
beginning to be deployed to support grid planners in optimise grid systems to take
Scaling positive impact thanks to AI more rapidly evaluating potential points of connection
AI has the potential to exponentially scale this and solutions. This is increasingly important given advantage of greater interactivity
approach beyond buildings to infrastructure and
vehicles to enable planning of demand coupled with
that many utility grids are rapidly ageing and need to
adapt quickly to meet the needs of the future.
and more complex and
distributed supply and storage across entire cities or intermittent sources of demand
regions. The possibility of a carbon-free economy
With development and use of the right AI tools, in 20 and supply.”
Demand modelling assumes perfect foreknowledge years we could live in a world where the electricity
of the weather and conditions on the grid—conditions and green molecular fuels that support an entirely
which are never met in operation. AI, coupled with carbon-free economy are dispersed throughout the
sensors deployed in building thermostats, vehicles, energy grid, right-sized to meet needs without over-
solar panels, storage, and other distributed devices building, and operated across complex systems
can effectively manage grid demand far more respecting end-users’ needs, vehicle charging and
responsively than a human operator. Learning from transportation requirements, and the intermittency

13
AI for Future Cities 02: Energy

Case Studies:
What is Arup currently doing to make cities better through AI?
Arup has worked with partners and
Arup Mass Energy
clients around the world on AI-driven United States of America, 2019 – present
Read more

solutions for real-world application that Creating energy models is a time-consuming process, and the
The Mass Energy tool has been developed on several projects
over recent years, including Colegio Village, NYSERDA, New
deliver better outcomes for cities. effort involved in modifying these models often significantly
York Public Libraries, City of Boston, and Whole Foods to run
constrains the breadth of options that can be studied. Arup’s
multiple EnergyPlus models programmatically that simulate
Mass Energy tool provided an automated, data-driven solution
These cutting-edge projects give us to this constraint. The tool leverages proprietary, AI-driven
and measure energy consumption, carbon emissions, and other
metrics.
powerful insight into how AI may genetic algorithms to rapidly analyse the vast spectrum of design
options and identify thousands of different ways to achieve
shape the future expectations of project sustainability targets. Options are presented through an
practitioners and how these may be interactive interface that enables easy filtering and comparison.
Among the many options identified that meet sustainability
transferable to other domains across the targets, project teams can explore and select the options that are
city. What if all buildings and projects across the city optimized for other project criteria such as cost or conformance
were optimised to maximise building performance? to preferred building systems.

ELEMENTS, MTR Read more

Hong Kong, 2025
Additional examples of where Arup have leveraged feedback, while reinforcement learning allows the system to
Located above Kowloon Station and the Airport Express,
digital technologies to achieve remarkable outcomes in adapt its operations based on performance outcomes.
ELEMENTS, an Arup and Neuron collaboration, is a one million
energy projects can be found here:
square foot mall combining shopping, dining, an ice rink and Recognised for its integration of advanced engineering and
cultural attractions. To enhance energy performance and thermal intelligent technology, the project received both the ASHRAE
SCALE Read more comfort, the Neuron AI Chiller Plant Platform was implemented, Engineering Excellence Award and the Technology Award
Offshore wind deployment resulting in a 9% energy saving in air conditioning and in 2025, marking a notable milestone in data-driven energy
Carbon free Boston mechanical ventilation (ACMV) systems within the first year, optimisation within East Asia’s built environment.
Read more
United States of America with an expected 15% following planned system upgrades.

One Taikoo Place Using Recurrent Neural Networks and Long Short-Term
Read more Memory networks, the Neuron AI model predicts cooling loads
Hong Kong
What if every building had an AI brain allowing in 30-minute intervals, drawing on real-time weather forecasts,
Flexibility Markets Unlocked solar radiation and historical data. A digital twin enables
Programme Read more it to communicate and share resources with other
buildings across the city? continuous AI training through scenario testing and operator
United Kingdom

14
AI for Future Cities 02: Energy

Recommendations:
Practical steps for today’s practitioners
AI will impact energy in cities in profound ways.
Although we can’t predict these impacts exactly, we can
1 Support regulation to enhance data 3 Invest in training, upskilling and
sharing and interoperability at scale raising public awareness
take anticipatory steps today to be better prepared for Government and regulators should enhance data sharing AI will introduce new data, actors and sectors into the
the future city. and interoperability through enabling a common data energy grid that have historically been regarded as
sharing infrastructure, providing clarity for market actors. unrelated – including everything from meteorology to
These recommendations seek to provide the
The energy trilemma of balancing security, cost and demography to high-street footfall. This will require
foundational elements and enablers that are required
sustainability should be at the heart of this. In the process greater public awareness of the implications of AI as well
for AI fully to augment our energy system at scale, of making these reforms, regulators, utilities and grid as a greater need to invest in the skills to take advantage
including interoperability, data quality, accessibility, operators must prepare backup plans for new security of AI. Organisations will need to invest in upskilling of
regulatory standards and system resilience. challenges of a connected grid with diverse actors, practitioners, who will increasingly need a deep systems-
complex systems and black boxing. level understanding of energy networks, the remit of AI
applications and the governance and ethical challenges
associated with AI assisted decision making.
2 Establish trust frameworks and transparent
data sharing agreements
For interoperability across an AI-enabled energy system to
be successful at scale, all actors across a more distributed 4 Leverage distributed energy and greater insight
urban energy value chain should have full visibility in for local need
data sharing agreements, frameworks, and standards. City and municipal leaders should explore opportunities to
This should involve creating shared key concepts proactively engage with the evolving AI-enabled energy
(ontologies & taxonomies etc.) as well as common system, which is becoming more open-source, distributed,
exchange mechanisms and data sharing infrastructures and automated. As urban energy data becomes richer
to enable AI to more seamlessly integrate across the and more interoperable, new possibilities will arise to
network and reduce data hoarding from stakeholders. understand local needs and develop innovative projects,
tools, and initiatives that address acute community energy
challenges.

15
AI for Future Cities 02: Energy

5 Move beyond the asset 7 Design for consumer and prosumer

AI is already providing valuable insights in how we accessibility
manage resource use and operation at an asset and From charging vehicles, to using heat pumps and solar
building level, but a shift is needed to achieve the panels in urban homes, an AI-enabled urban energy
promised benefits of AI on a broader scale. That shift system will require a change in behaviours and involve
is to move from building-scale planning to larger-scale transitional disruption for users. The roll-out of AI,
planning, that looks at a neighbourhood, district, or city therefore, should be easy to learn and user-centric, with
scale through initiatives such as Local Area Energy systems designed to maximise user benefits transparently
Planning (LAEPs) and Regional Energy Strategic Plans and with minimal input. Stakeholders across the energy
(RESPs). system, should use AI to support breaking down barriers
to adoption while also giving users greater agency around
how they consume, generate, and sell their energy.
6 Build in resilience across a more distributed
network
As we utilise AI to support the transition away from
a centralised energy infrastructure to one that’s more
distributed, regulatory bodies, utilities and municipalities
should take steps to build new forms of redundancy and
resilience, and strengthen network connections. This could
include policies which mandate redundant infrastructure,
development of right-time monitoring systems, self-
healing grids and enhanced energy storage systems that
provide backup during disruptions.

16
AI for Future Cities 02: Energy

Get in touch with our experts to discuss further.

Arup University is our corporate university

function that prepares our clients, collaborators,
and global membership for the future. We are
committed to raising technical standards across
the natural and built environment sectors.
17
References
Bosman, F. (2024, February 9). New AI tool discovers realistic ‘metamaterials’
with unusual properties. Phys.org. https://phys.org/news/2024-02-ai-tool-realistic-
metamaterials-unusual.html Get in touch with our experts to discuss further.

Calma, J. (2024, February 7). How virtual power plants are shaping tomorrow’s Contacts
energy system. MIT Technology Review. https://www.technologyreview. Arup University Foresight
com/2024/02/07/1087836/how-virtual-power-plants-are-shaping-tomorrows-energy- [email protected]
system/
Arup Energy
Castelvecchi, D. (2023, November 22). AI is learning how to improve the power grid. www.arup.com/markets/energy/
MIT Technology Review. https://www.technologyreview.com/2023/11/22/1083792/ai-
power-grid-improvement/ Arup Cities
www.arup.com/markets/cities/
International Energy Agency. (2023, November 2). Why AI and energy are the new
power couple. https://www.iea.org/commentaries/why-ai-and-energy-are-the-new- AI for Future Cities Team
power-couple Arlind Neziri
George Harrington
Liu, L., Guo, X., Liu, W., & Lee, C. (2021). Recent Progress in the Energy Harvesting Julien Clin
Technology—From Self-Powered Sensors to Self-Sustained IoT, and New Applications. Mona Ivinskis
Nanomaterials, 11(11), 2975. https://doi.org/10.3390/nano11112975 Ness Lafoy
Sarah Bushnell
McKinsey & Company. (2024, October 24). AI power: Expanding data center capacity Tobias Revell
to meet growing demand. https://www.mckinsey.com/industries/technology-media-and-
telecommunications/our-insights/ai-power-expanding-data-center-capacity-to-meet- Arup Expert Contributors
growing-demand Giuseppe Di Stefano
Henry Brooke
National Engineering Policy Centre. (2025). Engineering Responsible AI: foundations Jemima Bruin-Bland
for environmentally sustainable AI. https://nepc.raeng.org.uk/media/2aggau2j/ Jhan Chan
foundations-for-sustainable-ai-nepc-report.pdf Poya Rasekhi
Rob Best
SemiAnalysis. (2024, March 13). AI datacenter energy dilemma race. SemiAnalysis. Simon Evans
https://semianalysis.com/2024/03/13/ai-datacenter-energy-dilemma-race/#semianalysis- Will Cavendish
blocks-login-to-view-content
External Expert Contributors
World Nuclear News. (2025, January 13). UK government considering role for SMRs Lyndon Ruff, NESO
in AI expansion. World Nuclear News. https://www.world-nuclear-news.org/articles/uk- Magnus Cormack, NESO
government-considering-role-for-smrs-in-ai-expansion

Zhou, Y., Yang, Y., Yang, H., & Zhang, X. (2025). AI for energy: A review,
reassessment, and projection. Applied Energy, 354, 121942. https://doi.org/10.1016/j.
apenergy.2024.121942