The widespread adoption of AI technologies has drastically increased data processing demands, driving the expansion of data centers and the need for advanced infrastructure. However, energy requirements are rising exponentially, highlighting the critical need for renewable energy sources, efficient cooling systems, and optimized designs. This article explores the rapid growth of AI applications and its impact on data center infrastructure, energy consumption, and sustainability efforts worldwide. Additionally, it discusses the environmental challenges posed by energy-intensive AI training, the competitive landscape among nations to establish sustainable data hubs, and emerging regulations aimed at reducing emissions. Countries with favorable conditions—like affordable power, modern infrastructure, and regulatory support—are seizing the opportunity to establish themselves as leaders in AI-driven data infrastructure.
1. Growth in AI Applications and Infrastructure
The rapid growth of consumer AI adoption, led by OpenAI, has driven many large technology companies to accelerate their shift toward large language models and other AI technologies to provide innovative solutions and remain competitive in both the public and private sectors. As of 2024, there are approximately 70,000 AI companies, with the highest surge in acquisitions occurring in 2019, when 242 companies were acquired.[1] The expansion of AI applications—ranging from natural language processing (e.g., ChatGPT) and computer vision to autonomous systems—has drastically increased data processing demands. The volume of data generated, captured, copied, and consumed is projected to grow from 41 to 147 zettabytes, pressuring existing data centers to meet soaring requirements.[2] In response, the number of data centers worldwide has not grown as significantly, increasing only from around 7,000 to 8,000 during the same period, with the majority still concentrated in the U.S., Germany, the UK and China.[3] Meanwhile, competition to lead in developing new server hubs has intensified, both within the tech sector and among nations striving to dominate in data infrastructure. India, Italy, Colombia, Malaysia and Saudi Arabia are some of the few regions where cloud providers already agreed to expand.[4] Although around 40 countries have introduced local data storage mandates, especially to store financial and health data, most server hubs continue to operate across borders.[5]
AI-related server ecosystem involves many international operations such as high-processing chips, telecommunication networks, energy infrastructure, and cooling systems. Additionally, this ecosystem relies on advanced storage solutions, data management software, cloud operator presence, cybersecurity frameworks, and regulatory compliance with data protection laws across different countries. Considering the high processing power chips are coming from Taiwan for top cloud service providers such as Microsoft (OpenAI utilizes also Azure), Amazon, Meta and Google, there is an intense shift from China to a smaller country like Taiwan due to TSMC (Taiwan Semiconductor Manufacturing Company) that holds the expertise in advanced chips used in servers and AI applications worldwide. AI-related hardware exports of Taiwan surpassed China as of 2023, including liquid cooling systems that are required for most high-processing power server hubs.[6] Telecommunication companies in the U.S., UK, Japan, India, Singapore and UAE have strategic roles in undersea cable infrastructure, data centers, and advanced telecommunications networks.
2. Energy Consumption by Data Centers
Data centers and communication networks are already responsible for 2-3% of worldwide electricity use and contribute to 1% of global greenhouse gas (GHG) emissions.[7] As the need for higher processing power is needed, the energy need will be growing exponentially transitioning from CPUs (Central Processing Units) to GPUs (Graphics Processing Units) that can process larger amounts of data in higher speed computations with low latency and TPUs (Tensor Processing Units) that are more efficient in processing neural networks and machine learning applications. Countries that already have cloud server hubs must be prepared for this high-efficiency demand not only by capacity but also by providing a reliable energy grid infrastructure.
According to Goldman Sachs’ estimation, processing a ChatGPT query requires nearly ten times the electricity compared to handling a Google search.[8] Data centers are estimated to consume 463 TWh in 2024, a jump from around 200 TWh back in 2019.[9] Current energy infrastructure so far has been handling this increment; however, there must be new facilities or newly built expansion to the existing facilities. Certain server hubs are achieving impressive efficiency relative to their power requirements, even amid rising processing demands. As of 2024, the average Power Usage Effectiveness (PUE) ratio for data centers stands at approximately 1.56, indicating that for every watt used by IT equipment, an additional 0.56 watts are consumed by supporting infrastructure like cooling and power distribution.[10] This metric has remained relatively stable over the past five years.[11] However, newer and larger data centers are achieving lower PUE values, often around 1.3 or better, due to advancements in design and technology. As these modern facilities become more prevalent, the overall industry average PUE is expected to decline in the coming years. Generative AI has yet to become the primary workload in most data centers and is still far from reaching its peak; instead, these facilities predominantly manage high-performance computing tasks and business applications like in-memory transaction processing.
This shift opens an opportunity for countries beyond the U.S. and China to enter the competitive arena and establish themselves as leading regions. Unlike natural resources such as oil, which are geographically limited, server hubs can technically be built almost anywhere, enabling a broader range of regions to participate in the data center industry based on certain criteria such as land availability, sufficiency in electric grid infrastructure, cost of electricity, power capacity, risk factors, lower carbon emissions and tax benefits. Looking at countries’ electric grid age, the U.S. and Europe have the oldest infrastructure with over 40 years on average. This means a significant investment in the transmission and distribution of electricity must be implemented. Affordable power costs are crucial, and currently, Doha, Qatar; Quincy, Washington; and Vancouver, Canada, provide some of the lowest electricity rates available.[12] Server hubs should be in areas with minimal risk of natural disasters, avoiding regions prone to flooding, seismic activity, or hurricanes.
3. Cooling and Operational Demands
Higher energy consumption per server rack results in increased heat, driving the need for advanced cooling systems in high-performance data centers. Traditionally, air-cooled chillers have been preferred because they avoid water evaporation and can operate in regions with limited water supplies or extreme weather. However, for rack densities exceeding 80 kW, liquid cooling becomes necessary, and densities beyond 200 kW are expected in the future. Hybrid cooling systems combining air and liquid cooling, as well as fully liquid cooling, are already being implemented in some cases. This shift in cooling technology influences the choice of data center locations as a reliable water supply becomes essential. Regions with colder climates, like Nordic countries in the EU, offer an advantage by reducing the demand for additional cooling systems as well.
Coolants are also a major source of energy consumption accounting for 25% to 40% of total energy use in data centers.[13] Improving cooling system efficiency and adopting innovations are key to the operational advancement of AI data centers. Heated liquid from these systems can also be repurposed for various applications such as heating nearby buildings, supporting agricultural greenhouses, generating steam for industrial processes, or even supplying heat for district heating systems.
4. The Environmental Impact of AI’s Energy Needs
Despite significant strides toward renewable energy, coal remains a trusted and reliable power source for data centers. The EU and Latin America are well-positioned to provide renewable energy for data centers, with the EU sourcing 44.7% and Latin America & Caribbean sourcing 64% of its electricity from renewables in 2023, compared to about 22% in the U.S.[14] [15] [16] This strong reliance on clean energy gives Latin America and Europe an advantage in meeting data centers’ sustainability demands compared to other regions.
Hydropower has been utilized as a more stable source of renewables such as in Norway, in the U.S., in Canada to feed data centers and in Paraguay, particularly for Bitcoin mining operations. Google is exploring geothermal solutions, focusing on areas where natural geothermal conditions exist, such as hot water reservoirs or heat trapped between rock layers. Solar and wind are highly popular among leading cloud technology companies, with solar energy particularly favored for pairing with battery storage systems.
The intermittent nature of renewables, combined with scalability challenges and financial constraints during their development, has led data centers to explore alternative solutions. Nuclear power, particularly through Small Modular Reactors (SMRs), is emerging as a dependable option. Companies like Oracle, Google in collaboration with Kairos Power, AWS in partnership with X-Energy, and Deep Atomic in Switzerland have announced plans to utilize SMRs to supply electricity for their data centers, promising a stable, low-carbon energy source for these high-demand facilities. In Europe, SMRs are under development in Romania, Czechia, Poland, the UK, France, and Finland, while in the Middle East, the UAE, Saudi Arabia, and Qatar are planning SMR deployments. This movement is expected to facilitate the establishment of additional data centers or edge hubs in these regions. Notably, Poland, Romania, and the UK are repurposing decommissioned coal plants, transforming them into SMR-powered facilities.
5. Rising Demand for AI in Key Industries
The surge in e-commerce and online streaming since the COVID-19 pandemic has been impacting AI solutions to be developed faster, and media and entertainment is predicted to be 35% of the use of data centers.[17] Localizing online streaming and expanding e-commerce solutions will drive the need for more distributed edge solutions and server hubs, particularly within city centers. Operating these hubs in constrained spaces with high land costs presents challenges necessitating prioritization studies and careful research into the utilization of byproducts to optimize resources. Localized data centers are already mandating healthcare information, telecommunications data, financial and payment processing records, insurance data, and satellite or mapping information.
AI is streamlining IT processes, enabling rapid analytics and forecasting, detecting fraud, and enhancing customer and guest experiences across diverse sectors, particularly in finance and insurance, wholesale and retail, transportation and logistics, healthcare, and manufacturing. These industries, recognized as early adopters of AI, have already committed to further investments to explore AI applications across various operations, aiming to boost everyday work efficiency. Deployment of AI in its industries in China surpassed 58%, followed by India and Italy based on research done in 2022.[18]
Leading companies in AI patents include prominent tech giants like Baidu, IBM, and Microsoft, as well as State Grid Corp, a Chinese state-owned electric utility and the largest utility company globally.[19] This highlights an important use case of AI in predicting energy, resource needs, and distribution efficiency. These pioneering technology, energy, e-commerce and insurance companies set a precedent likely to be followed by others in different regions.
6. Government Regulations and Incentives
Certain regions offer sales and value-added tax benefits for data centers, along with others looking to reduce these taxes further. In the U.S., states like Oregon, Virginia, and Texas provide tax incentives on data center equipment and electricity use. Ireland and Finland are also leaders in offering corporate tax breaks, with Finland’s cool climate additionally helping to reduce operational costs. In the APAC region, Singapore and Malaysia’s Multimedia Super Corridor (MSC) provide attractive tax incentives for data centers. However, many of these benefits come with eligibility criteria, meaning smaller data centers may not qualify for such exemptions.
The EU’s Multiannual Financial Framework for 2021-2027 is set to enhance digital capacity through specific programs, particularly the Digital Europe Programme (DEP), the Connecting Europe Facility (CEF2), Horizon Europe, and the Space Programme. Within the DEP, €2.5 billion out of a €7.6 billion budget is allocated for AI.[20] The digital strategy focuses on increasing investment in AI research, innovation, and adoption, aiming to reach €20 billion in annual AI-related funding from both public and private sectors beyond 2020. Nordic countries have implemented carbon taxes, ranging from US$31 to US$137 per metric ton of CO₂, primarily targeting heating, with some exemptions for electricity production. Given the region’s suitability for Generative AI data centers, renewable energy sourcing is becoming essential to align with sustainability goals and regulatory requirements.
Recently, bipartisan U.S. senators urged Congress to approve US$32 billion in funding for AI research and development to counter China’s US$50 billion investment in the field.[21] The Partnership for Global Inclusivity on AI (PGIAI) brings together the U.S. Department of State and top tech firms—Amazon, Anthropic, Google, IBM, Meta, Microsoft, Nvidia, and OpenAI. With a combined commitment of over US$100 million, they aim to leverage their collective knowledge, resources, and connections to tap into AI’s potential for advancing sustainable development and improving the quality of life in developing countries.[22] This could help developing countries to bring advanced technology knowledge and operations to their countries.
The UAE, Saudi Arabia, and Qatar have each introduced National AI Strategies, aiming to drive advancements in AI and position themselves as leaders in the field. Google has launched its largest AI initiative in the MENA region, committing US$15 million by 2027 to enhance AI skills, fund research, and deploy AI-driven tools.[23] It also supports AI research in healthcare and climate through local grants and partnerships, including a cloud infrastructure collaboration in Saudi Arabia.
Singapore has become a key hub for sustainable data centers in Asia, establishing new standards to help data centers reduce energy consumption for cooling in hot climates. The government has recently introduced guidelines allowing them to operate safely at higher temperatures. By increasing temperatures by just 1°C, data centers can achieve a 2-5% reduction in cooling costs.
7. Commitment to Carbon-Free Data Centers and Renewable Energy Initiatives in Tech
Google has set a goal to achieve 100% carbon-free energy for its data centers by 2030. To support this target, it signed an agreement with Eneco to power its Netherlands data center with wind energy over the next decade.[24] Another strategy is to reduce power consumption during peak hours in the EU to ease grid pressure, a practice also implemented in Taiwan, particularly during the summer, due to its island grid structure. The use of AI to manage carbon emissions is not new; since 2016, Google’s DeepMind has been employed to predict and monitor heat fluctuations, optimizing the use of pumps, chillers, and cooling towers to improve accuracy in reducing carbon emissions.
Recognized as the world’s largest solar-powered data center, the Mohammed bin Rashid Al Maktoum Solar Park in Dubai, UAE, with a capacity of 100 MW, attracts companies like Dell, Huawei, and Microsoft.
Not only cloud technology companies but also mega data center providers Iron Mountain, Digital Realty Trust, QTS Realty Trust Inc, and Switch have collectively increased their renewable energy investments to a total of 1,500 MW. These investments are distributed across various locations such as Nevada, Georgia, and California.
Amazon Web Services (AWS) as the market leader in cloud services, has invested in over 500 solar and wind projects across 27 countries worldwide, with more than half consisting of on-site rooftop solar installations.[25] Microsoft has already been contracted for about 20 GW renewables, most of which will take years to deploy and is now building Stargate, a 5 GW AI data center, along with Open AI and possibly considering a nuclear power location.
AI for Earth is a Microsoft initiative dedicated to harnessing the power of AI to address critical environmental challenges in climate change, agriculture, biodiversity, and water. This initiative has supported more than 700 organizations globally, promoting the use of advanced technology and cloud services in sustainability practices. These efforts are expected to create a full-circle benefit, driving impactful environmental solutions worldwide.
8. Conclusion
As data centers increasingly rely on high-processing AI technologies, including GPUs and TPUs, their energy needs continue to soar, creating new challenges for grid reliability and carbon emissions. Investments in renewable energy sources such as wind, solar, and potentially nuclear power for data centers demonstrate a commitment to reducing the environmental impact of this growth. Meanwhile, regulatory efforts and incentives across the U.S., Europe, and APAC regions underscore the role of government in fostering sustainable practices and supporting regional AI ecosystems. This rapid evolution of AI-driven data centers also presents a significant opportunity for a broader range of countries to establish themselves as key players in the global data infrastructure landscape. With the flexibility to build server hubs in diverse locations, countries beyond traditional tech hubs can capitalize on this trend by offering favorable conditions such as affordable power, modern electric grids, and tax incentives. Emerging regions with cooler climates, low disaster risk, and growing tech ecosystems—such as Nordic countries, Southeast Asia, and parts of the Middle East—are well-positioned to attract data center investments. By seizing this moment, emerging economies can not only boost their digital infrastructure but also create jobs, stimulate technological advancement, and establish themselves as competitive destinations in the fast-expanding AI and data center industry.
[1] Exploding Topics, “How Many AI Companies Are There?,” 2024, https://explodingtopics.com/blog/number-ai-companies.
[2] World Economic Forum, “Data volume is soaring. Here’s how the ICT sector can sustainably handle the surge,” May 2, 2024, https://www.weforum.org/stories/2024/05/data-growth-drives-ict-energy-innovation/.
[3] “Overview,” Cloudscene, 2024, https://explore.cloudscene.com/for-buyers/.
[4] Cushman & Wakefield, “Global Data Center Market Comparison,” 2024, https://cushwake.cld.bz/2024-Global-Data-Center-Market-Comparison/22/
[5] OECD, “The Nature, Evolution and Potential Implications of Data Localisation Measures,” 2023, https://www.oecd.org/en/publications/the-nature-evolution-and-potential-implications-of-data-localisation-measures_179f718a-en.html.
[6] “AI’s $1.3 Trillion Future Increasingly Hinges on Taiwan,” Bloomberg, October 31, 2024, https://www.bloomberg.com/news/articles/2024-10-30/how-ai-s-1-3-trillion-future-increasingly-hinges-on-taiwan
[7] World Economic Forum, “Data volume is soaring. Here’s how the ICT sector can sustainably handle the surge.”
[8] “AI is poised to drive 160% increase in data center power demand,” Goldman Sachs, 2024, https://www.goldmansachs.com/insights/articles/AI-poised-to-drive-160-increase-in-power-demand.
[10] “Data center average annual Power Usage Effectiveness (PUE) worldwide from 2007 to 2024,” Statista, 2024, https://www.statista.com/statistics/1229367/data-center-average-annual-pue-worldwide/.
[11] “Uptime Institute Global Data Center Survey 2024,” Uptime Intelligence, 2024, https://datacenter.uptimeinstitute.com/rs/711-RIA-145/images/2024.GlobalDataCenterSurvey.Report.pdf?version=0.
[12] Cushman & Wakefield, “Global Data Center Market Comparison.”
[13] “The Untapped Potential of Heat Reuse in Data Center Cooling Design,” Data Center Frontier, 2024, https://www.datacenterfrontier.com/sponsored/article/55089191/trane-the-untapped-potential-of-heat-reuse-in-data-center-cooling-design.
[14] “Latin America and Caribbean,” Ember, 2024, https://ember-energy.org/countries-and-regions/latin-america-and-caribbean/#:~:text=Latin%20America%20and%20the%20Caribbean,of%20the%20region’s%20clean%20power.
[15] “Renewables take the lead in power generation in 2023,” Eurostat, 2024, https://ec.europa.eu/eurostat/web/products-eurostat-news/w/ddn-20240627-1.
[16] EIA, “Solar and wind to lead growth of U.S. power generation for the next two years,” 2024, https://www.eia.gov/todayinenergy/detail.php?id=61242.
[17] “Data Center Market Outlook from 2024 to 2034,” Future Market Insights, 2024, https://www.futuremarketinsights.com/reports/data-center-market.
[18] “Rate of adoption and deployment of artificial intelligence (AI) in enterprise globally and in selected countries in 2022,” IBM, 2022, https://www.statista.com/statistics/1378695/ai-adoption-rate-selected-countries/.
[19] LexisNexis & Patent Sight, “Largest patent owners in machine learning and artificial intelligence (AI) worldwide from 2013 to 2022, by number of active patent families,” 2022, https://www.statista.com/statistics/1032627/worldwide-machine-learning-and-ai-patent-owners-trend/
[20] European Parliament, “The Role of Artificial Intelligence in the European Green Deal,” 2021, https://www.europarl.europa.eu/RegData/etudes/STUD/2021/662906/IPOL_STU(2021)662906_EN.pdf.
[21] “US Lawmakers seek $32 billion to keep American AI ahead of China,” Reuters, 2024, https://www.reuters.com/world/us/us-senators-unveil-ai-policy-roadmap-seek-government-funding-boost-2024-05-15/.
[22] US Department of State, “United States and Eight Companies Launch the Partnership for Global Inclusivity on AI,” 2024, https://www.state.gov/united-states-and-eight-companies-launch-the-partnership-for-global-inclusivity-on-ai/.
[23] “Google to Invest $15M in AI Skills, Research, and Infrastructure for MENA,” Edge, 2024, https://www.edgemiddleeast.com/ai/google-to-invest-15m-in-ai-skills-research-and-infrastructure-for-mena.
[24] “Google and Eneco working together to make Eemshaven data centre power more sustainable,” Eneco, 2023, https://news.eneco.com/google-and-eneco-working-together-to-make-eemshaven-data-centre-power-more-sustainable/.
[25] “Amazon: All our operations now run on renewable energy,” Data Center Dynamics, 2024, https://www.datacenterdynamics.com/en/news/amazon-all-our-operations-now-run-on-renewable-energy/.
