Why Does Artificial Intelligence Depend on Critical Minerals?
Introduction
Data centers consumed 415 terawatt-hours (TWh) of electricity in 2024, representing approximately 1.5% of global electricity consumption [1]. This energy demand could more than double to between 620 and 1,000 TWh by 2030, an increase largely attributable to the rapid expansion of AI (artificial intelligence). The spectacular rise of artificial intelligence is not solely based on software or algorithmic advances. It relies on a complex material infrastructure, invisible to the user, but essential to the functioning of contemporary digital technologies.
Behind this apparent dematerialization lies a profoundly material reality: critical minerals. Although AI is perceived as an immaterial and digital technology, it depends entirely on physical resources extracted from the Earth. This article explores the critical minerals essential to AI, the physical infrastructure required for artificial intelligence technologies, and the concrete applications using these minerals.
The Explosive Rise of AI and Its Energy Needs
The artificial intelligence revolution is measured not only in algorithms and lines of code, but also in gigawatts of electrical power and tonnes of materials. Data centers, the infrastructure at the heart of AI, are experiencing unprecedented growth that is radically transforming the global energy landscape.
From 2024 to 2030, the electricity consumption of data centers will experience double-digit annual growth, significantly faster than the growth in total electricity consumption of other sectors [1]. This spectacular progression reflects the intensification of workloads related to artificial intelligence. To put these figures in perspective, the projected increase from 415 TWh to 945 TWh represents a rise of 127% in just six years, a growth trajectory that surpasses that of most industrial sectors.
Energy-Hungry Equipment
The energy intensity of AI equipment represents a major challenge. A single graphics processing unit (GPU), such as the Nvidia H100, can consume more than 700 watts [3]. To illustrate this demand, a data center operating 100,000 of these GPUs generates gigawatts of computing power and requires a substantial volume of critical minerals to power the infrastructure and maintain cooling operations [3]. These massive energy needs explain why AI data centers require considerably more robust material infrastructure than their predecessors.
Which Critical Minerals Are Indispensable to AI
Behind every AI chip and every server lies an arsenal of minerals with sometimes unfamiliar names, but essential properties. These materials form the physical foundation of the digital revolution.
Semiconductors, the true brains of AI systems, comprise approximately 300 different materials, including rare earth elements and other critical minerals [5]. Among the most crucial components are cerium, europium, gadolinium, lanthanum, neodymium, praseodymium, scandium, terbium, and yttrium, as well as the critical minerals gallium and germanium [5].
Also read : Discover the list of critical minerals
Specific Functions of Minerals in AI Infrastructure
Each mineral plays a precise role in the technological ecosystem of AI. Gallium and germanium form the basis of compound semiconductors and fiber optic connections used in AI accelerators and data center interconnects [6]. Yttrium finds its use in coatings and laser materials that improve chip efficiency while reducing the energy required for cooling.
Europium enhances optoelectronic and laser materials, and also serves to dope compound semiconductors. Gadolinium improves heat resistance in GPUs (Graphics Processing Units) and AI accelerators subjected to high workloads [7]. As for lanthanum, it is used in high-dielectric-constant dielectrics to improve transistor insulation and energy efficiency.
The Material Infrastructure of AI: Data Centers and Semiconductors
Artificial intelligence relies on a resource-intensive physical infrastructure, where every component, from the smallest transistor to the largest electrical cable, incorporates critical minerals.
Copper, the Cornerstone of Data Centers
A study of Microsoft's data center in Chicago, whose construction cost $500 million, revealed that it used 2,177 tonnes of copper, or up to approximately 27 tonnes of copper per megawatt (MW) of applied power in this type of large-capacity installation [9]. This ratio reflects the resource intensity of these facilities.
Global data center capacity is expected to more than triple, rising from approximately 60 gigawatts (GW) in 2023 to a range of 171 to 219 GW by 2030, with an optimistic scenario reaching 298 GW [10]. Approximately 70% of this growth will come from facilities built specifically to support advanced AI workloads [10].
Unprecedented Power Densities
High-performance AI data centers are distinguished by their elevated power densities. New racks can draw up to 120 kilowatts (kW) each, compared to only 5 to 10 kW in standard data centers [10]. This spectacular increase requires considerably more robust electrical wiring and more sophisticated power distribution systems. Copper remains indispensable in power cables, busbars, cooling systems, circuit boards, and grounding systems [9].
Energy Storage Systems and Backup Batteries
Data centers require uninterruptible power supply (UPS) systems that incorporate backup batteries to ensure continuity of operations [3]. Lithium, cobalt, and nickel are the essential components of battery chemistry. The lithium-ion battery market for data centers reflects this growing demand and is expected to reach $17.69 billion USD by 2034 [11].
Materials for AI Chip Production
The manufacturing of AI chips itself requires a variety of specialized materials. Silicon serves as the base substrate for semiconductor wafers. Copper acts as the primary conductor in integrated circuits thanks to its excellent electrical conductivity. Cobalt, for its part, provides better resistance to electromigration and is increasingly used in advanced semiconductor nodes [12].
Quebec and Canada in the Race for Critical Minerals
Faced with this growing demand and geopolitical challenges, Quebec and Canada possess strategic assets to become major players in the supply of critical minerals.
Quebec has recognized 28 critical and strategic minerals, several of which were added in January 2024 and are specifically relevant to AI, including germanium, high-purity silica, and high-purity iron [4]. This official recognition reflects the Quebec government's commitment to developing a complete value chain in this strategic sector.
Conclusion
The artificial intelligence revolution cannot materialize without reliable access to critical minerals. Minerals such as gallium, germanium, copper, and rare earths are indispensable to every component of this emerging technology's infrastructure, from semiconductors to cooling systems and data transmission cables.
This fascinating paradox deserves reflection: the digital future depends on terrestrial resources. This reality underscores the importance of developing a responsible mining industry that combines technological innovation, environmental respect, and supply security.
The challenges ahead are numerous: competition with the energy transition for the same resources, geopolitical concentration of current production, and the need for innovation in extraction and recycling. However, these challenges also represent opportunities for jurisdictions that can develop complete and sustainable value chains.
Follow Squatex on LinkedIn to stay informed of developments in the critical minerals and renewable energy sector.
References
[1] International Energy Agency (IEA). "Energy Demand from AI – Energy and AI." IEA Analysis, 2025. https://www.iea.org/reports/energy-and-ai/energy-demand-from-ai
[3] Impossible Metals. "Powering AI: The Critical Mineral Demands of Emerging Data Centers." Impossible Metals Blog, 2024-2025. https://impossiblemetals.com/blog/powering-ai-the-critical-mineral-demands-of-emerging-data-centers/
[4] Gouvernement du Québec. "Minéraux critiques et stratégiques." Québec.ca, 2024. https://www.quebec.ca/agriculture-environnement-et-ressources-naturelles/mines/mineraux-substances-minerales/mineraux-critiques-et-strategiques
[5] Information Week. "Is AI Driving Demand for Rare Earth Elements and Other Materials?" Information Week, February 11, 2025. https://www.informationweek.com/machine-learning-ai/is-ai-driving-demand-for-rare-earth-elements-and-other-materials-
[6] Tech Journal UK. "Critical Minerals Become Essential to AI Chips and National Security." Tech Journal UK, December 17, 2025. https://www.techjournal.uk/p/critical-minerals-become-essential
[7] SFA Oxford. "Critical Minerals in Artificial Intelligence." SFA Oxford Knowledge and Insights, 2025. https://www.sfa-oxford.com/knowledge-and-insights/critical-minerals-in-low-carbon-and-future-technologies/critical-minerals-in-artificial-intelligence/
[8] FP Analytics. "Artificial Intelligence and the Critical Minerals Crunch." FP Analytics, October 27, 2025. https://fpanalytics.foreignpolicy.com/2025/07/18/artificial-intelligence-critical-minerals-supply-chains/
[9] BHP. "Why AI Tools and Data Centres are Driving Copper Demand." BHP Insights, January 2025. https://www.bhp.com/news/bhp-insights/2025/01/why-ai-tools-and-data-centres-are-driving-copper-demand
[10] Nicola Mining. "Copper: The Unseen Catalyst of AI and Energy Transition." Nicola Mining, 2024. https://nicolamining.com/copper-the-unseen-catalyst-of-ai-and-energy-transition-2/
[11] Precedence Research. "Data Center Lithium-Ion Battery Market." Precedence Research, 2024. https://www.precedenceresearch.com/data-center-lithium-ion-battery-market
[12] Sentisight AI. "Materials Required for AI Chip Production." Sentisight AI, 2024. https://www.sentisight.ai/materials-required-for-ai-chip-production/
