America is doubling down on artificial intelligence (AI), with ambitions of global leadership playing out in multi-billion-dollar investments like The Stargate Project. But powering the AI era is about more than just silicon chips and server farms. The AI boom is redefining the future of American infrastructure. From hyperscale cloud clusters to edge computing and generative AI, the demand for data processing power is growing exponentially, and so is the need for the materials that support it. Copper, in particular, is critical to this transformation.
As OEMs, developers, and political leaders invest in a more digitally enabled, AI-powered economy, the question is no longer whether there’s enough copper in the ground. The U.S. has it. The critical challenge is ensuring that copper is mined, refined, recycled, and traded in a way that meets escalating demand without delay or disruption.
AI Data Centers Are Redefining Copper Demand
The AI revolution is placing unprecedented demands on data center infrastructure. Compared to traditional facilities, these next-generation facilities consume more power, generate more heat, and require more sophisticated infrastructure to manage it all. And copper is central to nearly every one of those functions. It delivers electricity, handles thermal loads, and supports connectivity.
With power density in AI server racks continuing to rise and cooling demands escalating, copper’s electrical and thermal properties make it the material of choice. These data centers are not only growing in number but also in scale. Hyperscale facilities can occupy hundreds of thousands of square feet, driven by the need to house the vast computing power required for AI workloads. This immense scale amplifies the demand for copper in power distribution, cooling, and network connectivity.
A conventional data center might use between 5,000 to 15,000 tons of copper. In contrast, hyperscale AI data centers, such as those built to house Nvidia’s HGX systems, can use up to 50,000 tons of copper per facility.
This increase is driven by several factors:
- Power delivery and grounding: These facilities, which house vast arrays of servers and networking equipment, are inherently energy intensive. While smaller data centers have power demands in the range of 5 to 10 megawatts (MW), hyperscale data centers require extensively more, with the International Energy Agency (IEA) reporting power demands more than 100MW, including some now called Gigawatt (GW) AI factories. of 100 MW or more. This significant power demand makes copper's superior conductivity essential for the reliable and efficient distribution of power in data centers. It is critical for wiring, busbars, electrical connectors, power distribution strips, and cabling, minimizing energy loss and optimizing efficiency. Copper is also vital for grounding systems, protecting sensitive equipment from electrical surges and ensuring operational stability and safety. Copper remains crucial for power delivery, and its high load capacity is well-suited to the increasing power density of AI server racks, which are driving even greater energy consumption in some hyperscale facilities. And copper's conductivity and corrosion resistance make it ideal for grounding rods, wires, and busbars, protecting against electrical faults to ensure hardware longevity. This ensures the operational reliability and longevity of power infrastructure, which is paramount for uninterrupted AI operations.
- Thermal management: High-performance computing generates significant heat. Copper is widely used in liquid cooling systems, heat exchangers, and thermal interfaces to dissipate that heat efficiently. To effectively manage the heat generated by AI workloads, advanced cooling solutions are essential. Liquid cooling, which utilizes copper cold plates and heat exchangers, is becoming increasingly prevalent due to its superior heat transfer capabilities compared to traditional air cooling. Copper's superior heat dissipation maintains optimal temperatures and prevents equipment failure. While aluminum is an alternative, copper's thermal conductivity in compact spaces often justifies its higher cost in high-performance AI environments, where even small temperature increases can harm efficiency and lifespan. Copper's recyclability also aligns with the sustainability goals of modern data centers.
- Data transmission: High-speed interconnects and networking equipment in AI clusters, especially in edge facilities where latency and reliability are key, rely on copper components. It is the only material used in Ethernet cables for short-distance, high-speed data transmission between servers and other equipment. While fiber optic cables are preferred for long-distance, high-bandwidth connections, copper remains a cost-effective and reliable solution for shorter connections, not requiring the cost and heat from lasers and delivering power to network devices. In fact, there’s over 2 miles of NVLink copper cables in massive bundles at the back of NVIDIA’s 120kW DGX GB200 NVL72 rack, which mesh the GPUs together to function collectively. This complementary use of copper and fiber meets the diverse connectivity needs of modern data centers. Copper is also vital for grounding systems, protecting sensitive equipment from electrical surges and ensuring operational stability and safety. Copper’s stable and high-speed data transfer is crucial for efficient AI training and inference, and copper's conductivity and corrosion resistance make it ideal for grounding rods, wires, and busbars, protecting against electrical faults to ensure hardware longevity.
To enable rapid deployment and scalability, prefabricated and modular data center designs are on the rise. These designs utilize copper for internal wiring, power distribution, and cooling systems within the modules, highlighting copper's critical role even in innovative data center architectures.
Demand from AI and data infrastructure is no longer theoretical. In 2024, North American data center construction reached record levels, with over 6.3 GW of capacity underway in primary US markets alone by the end of the year. This represents a substantial increase from the previous year and underscores the rapid expansion driven by the escalating demand for AI and digital services. While a significant portion of this development is geared towards meeting the intense computational needs of AI, the overall investment in data center construction continues to climb, with nearly $1.5 trillion announced recently in the U.S. alone, reflecting the critical importance of this infrastructure in the modern economy.
The U.S. Has the Copper, But Access Is Not Automatic
The U.S. Geological Survey (USGS) estimates 48 million metric tons of identified copper resources domestically. That’s well more than enough for decades to support this projected growth in AI infrastructure and the broader energy transition. However, those resources must be converted into usable metal, at speed and scale. That’s where the challenge lies.
While the U.S. was once a global leader in copper production, domestic mining output has declined by 6% over the past decade, and refining capacity is increasingly outdated. Today, more than one-third of the refined copper used in the U.S. is imported, often from countries with less transparent supply chains or geopolitical volatility. Permitting delays compound the issue. The time required to bring a new copper mine online in the U.S. averages 29 years, the second longest in the world, compared to less than five years in many other jurisdictions. This lag creates a mismatch between project timelines for critical infrastructure and material availability.
An All-of-the-Above Strategy Is the Path Forward
To secure copper supply for the AI era, the U.S. must adopt an integrated approach that combines:
- Increased Domestic Production: Streamlining federal and state permitting processes, providing regulatory certainty, and investing in brownfield expansion are key. For example, projects like Resolution Copper in Arizona and the re-opening of the White Pine mine in Michigan could meaningfully increase U.S. output, if supported by policy. This could involve setting clear timelines for permit reviews, offering financial incentives for sustainable mining practices, and supporting research into more efficient extraction technologies.
- Modernized Refining and Smelting: The U.S. has just three primary copper smelters, of which only two are operating, none of which are modern by international standards. Expanding and upgrading this infrastructure would improve domestic supply security and reduce dependence on imports. Investments in modern smelting technologies can increase efficiency, reduce emissions, and handle a wider range of copper concentrates.
- Increased Domestic Recycling: The copper recycling rate in the U.S. remains underutilized. Enhanced collection systems, incentives for OEM take-back programs, and investment in clean separation technologies could unlock millions of metric tons of secondary copper. Already, roughly one-third of U.S. copper consumption comes from recycled material, a figure that could grow substantially. This recycled copper also must stay and be used domestically. The U.S. is by far the largest exporter of copper scrap. Recently, four secondary smelters have opened, increasing secondary smelting capacity another 400,000 metric tons.
- Reliable Trade Partnerships: While domestic capability must be prioritized, copper remains a global market. Ensuring trade access with reliable trade allies such as Chile, Canada, Peru, and Mexico helps buffer short-term fluctuations and provides flexibility. This involves maintaining strong diplomatic relationships, negotiating favorable trade agreements, and diversifying import sources to mitigate risks.
The U.S. Has What It Takes to Keep Pace
At the Copper Development Association, we support a proactive, collaborative approach between government, OEMs, and the copper industry. The reserves exist. The industrial base exists. What’s needed now is urgency and alignment.
Copper is not a limiting factor for America’s AI future, unless we allow it to become one. By taking decisive action to secure a reliable copper supply chain, the U.S. can ensure its continued leadership in the AI revolution and strengthen its economic and national security for decades to come.