Artificial intelligence (AI) is rapidly gaining ground, revolutionizing industries and shaping the future of our digital world. As AI technologies become increasingly embedded in everyday life, the infrastructure supporting them, particularly data centers, is expanding quickly. These facilities are essential for powering AI, but they come with a significant downside: a growing appetite for energy. This surge in demand raises critical questions about how to power data centers sustainably while ensuring efficiency and reliability. In this article, we explore geothermal energy combined with Organic Rankine Cycle (ORC) technology as a promising low-carbon solution for generating the electricity required by data centers. We also look at how the substantial heat generated during data center cooling processes can be recovered and reused through heat pumps to supply urban district heating systems. By integrating these approaches, we can turn energy-intensive infrastructure into a key player in the transition toward more sustainable cities.
1. AI’s Rapid Rise and Transformative Potential
Artificial intelligence (AI) is rapidly transforming industries and everyday life, driven by breakthroughs in computing power, data availability, and innovative model architectures. AI systems are now widely adopted across sectors—from manufacturing and healthcare to logistics and energy.
One practical example is BrainBox AI, which applies machine learning to optimize industrial and commercial HVAC systems. By integrating data from building controls, occupancy, energy tariffs, weather forecasts, and carbon metrics, the system can autonomously adjust operations to enhance energy efficiency, comfort, and emissions reduction. This type of AI-driven optimization illustrates the broader promise of artificial intelligence: increasing performance while reducing waste across industrial ecosystems.
The energy sector is becoming more electrified, digitalized, and decentralized, creating both challenges and opportunities. AI solutions are proving essential to optimize operations, reduce costs, improve efficiency, and cut emissions—potentially unlocking savings of up to $110 billion annually and 175 gigawatts of transmission capacity, according to the IEA.
However, increased digitalization introduces new vulnerabilities. The rapid expansion of electricity demand, driven in part by AI data centers, intensifies pressure on supply chains for essential grid components and critical minerals like copper and gallium, raising concerns about global supply security.
Together, these dynamics highlight both the transformative potential of AI in energy and the complex challenges it brings—underscoring the need for sustainable infrastructure solutions.
2. Energy Overload: AI’s Impact on Efficiency and Compliance
The rapid expansion of AI workloads is driving a significant surge in energy demand, particularly within data centers where AI models are trained and deployed. These generative models require vast computational power continuously, which not only increases electricity consumption but also intensifies cooling requirements, pushing the Power Usage Effectiveness (PUE) metric higher. This means a growing portion of energy is devoted to supporting infrastructure rather than the core IT operations, challenging data centers to balance performance with efficiency.
Traditional data centers typically use between 10 and 25 megawatts (MWe) of electricity. However, next-generation hyperscale AI data centers can demand over 100 MWe each—equivalent to the annual electricity consumption of 100,000 households. One of the largest planned data centers could even consume as much electricity as 5 million homes, highlighting the scale of infrastructure being built to support AI development.
Current projections indicate that in Europe alone, energy consumption by data centers could jump from around 62 terawatt-hours (TWh) annually today to over 150 TWh by 2030, largely fueled by AI-related activities. This rise is expected to make data centers account for nearly 5% of Europe’s total power consumption within six years—more than double the current share. Globally, data centers represented about 1.5% of electricity use in 2024, with forecasts suggesting this could reach approximately 3% by 2030, amounting to nearly 945 TWh—roughly equal to Japan’s entire current electricity consumption.
The scale of this energy demand varies geographically, with data centers in the United States projected to contribute to almost half of the country’s electricity growth by 2030, and similarly high impacts anticipated in Japan and Malaysia. While a variety of energy sources will be employed to meet these needs, renewable energy and natural gas are projected to lead due to their availability and cost-effectiveness.
At the same time, regulatory pressures such as ESG commitments, Scope 2 emissions accounting (indirect emissions from purchased electricity), and the EU Green Deal are compelling data centers to adopt more sustainable and transparent practices. The focus is shifting from merely increasing energy supply to ensuring that power is intelligent, continuous, and clean. Solutions combining energy efficiency improvements, low-carbon energy sources, and innovative heat recovery systems will be essential to meet the escalating demands of AI-driven data centers without compromising environmental goals.
As AI continues to expand, so does the urgency to decarbonize the infrastructure that powers it. In the next chapters, we will explore how geothermal energy and industrial heat recovery can offer scalable and reliable solutions to align AI’s potential with global sustainability goals.
3. From geothermal to AI efficiency: ORC as a low-carbon solution for AI power
When it comes to powering AI infrastructure sustainably, not all renewables offer the same advantages. Geothermal energy, in tandem with Organic Rankine Cycle (ORC) technology, delivers a unique match for the demands of AI-driven data centers.
Geothermal:
- Alimentazione continua 24/7: a differenza delle fonti intermittenti (solare, eolico), la geotermia fornisce elettricità ininterrottamente—essenziale per i carichi di lavoro AI critici.
- Alta affidabilità e fornitura locale: riduce la dipendenza dalla stabilità della rete, evita colli di bottiglia nella trasmissione e mitiga i rischi di blackout.
- Bassa impronta di carbonio: perfetta per supportare gli obiettivi ESG e le strategie Net-Zero, fornendo energia pulita senza intermittenza.
ORC:
- Adapts to low- and medium-grade heat: ORC systems use organic working fluids to convert geothermal heat into electricity with thermal efficiencies typically between 3–7%, and up to 4–8% in optimized setups.
- Modular and scalable: ORC units can range from small applications to hyperscale installations, exactly matching data center sizes without oversizing.
- Low maintenance and high availability: with minimal moving parts and sealed systems, ORC units offer dependable performance and low operational costs.
Integration with AI Data Centers
- Right-size integration: ORC units can be installed on-site or nearby, reducing energy losses and avoiding grid congestion.
- Heat-to-power and heat-recovery synergy: ORC recovers waste heat and converts it to electricity, reducing overall energy usage.
A recent example of this shift is Google’s landmark power purchase agreement (PPA) with Baseload Power in Taiwan — the company’s first-ever geothermal PPA in Asia-Pacific, and Taiwan’s first corporate agreement of this kind. The deal will enable the addition of 10 megawatts of 24/7 geothermal power to the Taiwanese grid by 2029, effectively doubling the country’s current commercial geothermal capacity. In parallel, Google is also making a direct equity investment in Baseload Capital, Baseload Power’s parent company, to help scale geothermal infrastructure in the region. The clean electricity generated through this partnership will supply multiple local Google operations, including a data center, office facilities, and the company’s largest hardware R&D site outside the United States.
At Exergy, we develop and deliver advanced ORC systems specifically engineered for low-to-medium temperature geothermal applications.
Learn more about how our ORC technology can support your clean energy goals!
4. Heat recovery and heat pumps — converting data center waste heat into urban energy
AI data centers, in addition to their significant energy consumption, generate large amounts of waste heat from the cooling systems required to keep their infrastructure operational. This heat is often dissipated into the environment, but it can be a valuable resource—especially in urban areas where it can be captured and integrated into district heating networks.
In several Northern European cities, such as Stockholm and Copenhagen, data centers are increasingly recognized as important urban heat sources. Thanks to advanced heat recovery technologies, the heat produced is collected and distributed to thousands of homes via district heating systems, significantly reducing the reliance on fossil fuels for residential heating.
Heat pumps play a central role in this process: they extract low-grade heat generated by cooling systems, then upgrade the temperature to levels suitable for heating buildings or supporting local industrial processes. This reduces the demand on traditional heating infrastructure and creates direct environmental and economic benefits for local communities.
To fully unlock this potential, a keen approach to thermal management is essential, where cooling infrastructure components—chillers, pumps, fans—are managed as an interconnected system through advanced controls and real-time monitoring. Such integration optimizes heat capture and reuse while maintaining reliability and minimizing losses.
Beyond district heating, recovered heat can support local industrial processes and foster mutually beneficial energy partnerships, provided these synergies are embedded early in planning and development.
Exergy offers new X-HEAT line, a range of heat pumps specifically designed for recovering waste heat from industrial infrastructures including data centers. Exergy’s systems are highly efficient even at low temperatures, modular, and customizable to meet the demands of various settings.
Learn more about Exergy’s X-heat line of large heat pumps for heat recovery and how we can support your projects!
Conclusion
The rapid expansion of AI and data centers presents both challenges and opportunities for the energy transition. By embracing innovative solutions like geothermal power with ORC technology and effective heat recovery through heat pumps, we can build cleaner, more efficient infrastructures. Collaboration among technology providers, energy companies, and urban planners will be key to unlocking the full potential of these approaches. The time to act is now—designing scalable, sustainable systems today will shape not only the future of digital innovation but also our environmental legacy.
Looking to turn geothermal heat into clean energy? Discover our ORC solutions to generate reliable renewable power and drive sustainability.
Want to recover and reuse waste heat? Explore our heat pumps for efficient district heating integration.
For more information, contact us here. Let’s partner together!
Sources:
