Hello! I'm your friend, an economist and financial journalist. Today's article about "AI's consumption of copper" reveals a very interesting and often overlooked economic phenomenon: the industry that seems the most intangible—computing power—ultimately relies on the most tangible of materials: physical substances.
Many people think of AI as nothing but code, algorithms, and the cloud, and nothing related to soil or minerals. However, the faster AI operates, the more power grids and servers it consumes, and one of the key ingredients to feed these systems is copper.
Let me first summarize the main points in simple language, and then I'll break down this news article in detail from five different perspectives.
📝 Summary of Key Points
The logical chain in this article is clear: The explosion of AI → A surge in data centers → A sharp increase in power demand → A surge in copper demand → A shortage in copper supply.
In short, AI itself doesn't directly consume copper, but the data centers that power it require massive amounts of electricity, and the transmission, distribution, and internal connections of these servers all depend on copper. Data shows that by 2040, data centers will account for about 6% of the world's copper consumption. However, due to the long mining cycle of copper, declining ore quality, few new discoveries, and the competition from electric vehicles and power grid construction, global copper supply could fall 25% short of demand. Although recycling and alternative materials (like aluminum) can help, copper's status as the "blood of industry" has only been reinforced in the AI era.
---
🔍 In-Depth Analysis: The Relationship Between AI and Copper
1. Why Does AI Consume So Much Copper? Because It's the "Transporter" of Electricity
There's a common misconception that AI is purely software and doesn't need physical materials. But that's not the case. Think of AI as a supercomputer with a "brain" in the cloud and a network of "veins" (wires) all over its body to deliver energy.
- Electricity is AI's Fuel: According to the International Energy Agency (IEA), data centers used 415 billion kilowatt-hours of electricity in 2024, accounting for 1.5% of global electricity consumption. By 2030, this figure could double to 945 billion kilowatt-hours, equivalent to Japan's annual electricity use.
- Copper is the Wall of the Veins: Electricity transmission requires materials that are good conductors, corrosion-resistant, and can be drawn into thin wires, and copper is one of the best choices. From power plants to data centers to the servers inside, copper is everywhere.
- Data Proof: A recent study found that among the 20 essential minerals needed for AI data centers, copper makes up 83% of the total weight. In other words, most of the weight of a data center is copper. Just the facilities for transmitting and distributing electricity account for 64% of copper demand.
💡 Simplified Explanation: In the past, we said "computing power equals national strength." Now it's more accurate to say "electricity equals computing power, and copper equals electricity." Without copper, electricity can't be delivered, and no matter how intelligent AI is, it would "starve."
2. How Much Copper Does AI Really Consume? The Numbers Are Surprising
Let's get more specific.
- The Appetite of a Single Data Center: According to S&P Global, a large 230-megawatt AI training data center requires about 10,000 tons of copper. That's roughly enough to fill several hundred heavy trucks.
- Global Trends:
- 2025: Global data center copper consumption is expected to reach 1.1 million tons, accounting for 4% of total global copper consumption.
- 2040: This figure is projected to soar to 2.5 million tons, accounting for 6%.
- Exponential Growth: Global data center capacity will increase from 100 million kilowatts in 2022 to 550 million kilowatts by 2040. This means the AI industry will create a huge, growing demand for copper over the next 15 years.
💡 Simplified Explanation: If we consider global copper demand as a cake, data centers used to be a small part of it. But in the AI era, that part is rapidly expanding.
3. Is There Enough Copper? The Reality Is Challenging: Mining Is Getting More Difficult, and New Mines Are Scarce
Demand is soaring, but the supply side is facing major problems. It's like a restaurant that suddenly becomes very popular, but its supply chain is broken.
- Long Mining Cycle: It takes more than 10 years to go from discovering a mineral deposit to producing copper from a mine. So the current copper shortage is a result of decisions made decades ago.
- Declining Ore Quality: The IEA notes that since 1991, the average quality of global copper ore has decreased by about 40%. In other words, it now takes more copper to extract the same amount of copper.
- Fewer New Mines: Only 5% of the copper mines discovered in the past 35 years were found in the last decade. The easily exploitable copper mines on Earth are running out.
- Competitive Demand: AI isn't the only consumer of copper. Electric vehicles (EVs), traditional power grid upgrades, and renewable energy (wind, solar) are also competing for it.
💡 Simplified Explanation: Copper mines are like aging gold mines—both the amount of copper and the number of new mines are decreasing, while the demand from AI, EVs, and power grids is increasing. As a result, the supply-demand gap is widening. The IEA predicts that by 2035, global copper supply could fall 25% short of demand.
4. What If There's Not Enough Copper? Recycling Is the "Urban Mine," and Alternatives Are a Last Resort
To address the shortage, there are two main strategies: finding new sources and reducing consumption.
- Recycling: Turning Waste into Gold: Copper doesn't disappear when burned like coal; it just changes form. Copper from old wires and cables can be recycled with almost no loss in quality.
- Current Situation: Recycling accounts for 33%-37% of global copper demand. However, as new copper consumption grows rapidly, the proportion of recycled copper is declining.
- Future: If copper prices remain high, recycling will become more economical, and old appliances and buildings in cities could become important sources of copper.
- Alternatives: The Challenges of Aluminum and Silver:
- Aluminum: Although it's less conductive than copper, it's cheaper and lighter. It can replace copper in some applications where conductivity isn't critical.
- Silver: It's more conductive than copper but much more expensive, so it's only used in a few high-end applications.
- Technological Solutions: Using more efficient chips and transformers, or adopting direct-current power grids, can reduce copper usage.
💡 Simplified Explanation: Recycling is a long-term solution but won't solve the problem completely. Alternative materials have limitations in performance or cost. Copper's central role is unlikely to be replaced in the short term.
5. The Deeper Logic: Every Technological Revolution Redefines Basic Materials
The article concludes with the idea that the rise of copper is an inevitable result of technological revolutions.
- Historical Examples: Copper went from being a decorative material to a vital industrial asset in the 19th century, saw another surge in demand during the 20th century with telephones, home appliances, and the internet, and is now a key bottleneck in the 21st century with AI and electrification.
- Core Insight: As long as humans need to transmit electricity and information, copper will remain essential. AI doesn't create new physical laws, but it greatly increases the demand for existing infrastructure.
- Investment and Strategic Significance: For individuals and investors, copper is no longer just a commodity; it has become the "oil of the digital age." Companies that control copper resources, recycling technology, or energy-efficient technologies will play a crucial role in the coming decades.
💡 Simplified Explanation: Copper used to be considered a relic from the past, but now it's a valuable asset. The more advanced AI becomes, the more important copper becomes. This is a structural trend that will last for decades.
---
📌 Insights for Everyone
1. Don't Be Misled by the Virtual: AI, cloud computing, and big data may seem intangible, but they rely on physical materials like steel, copper, water, and electricity. Understanding this helps you better assess the true costs of technology industries.
2. Pay Attention to Resource Constraints: Future technological competitions may not be about who has the best algorithms but who has the most efficient energy use and a stable supply chain of materials.
3. Raise Awareness of Recycling: As copper prices may rise, the value of recycling old metals will increase. Old appliances and cables at home could become more valuable in the future.
4. Take a Long-Term View: The copper supply-demand gap is a long-term issue that won't be solved overnight. Any technology that claims to completely replace copper will struggle to be widely adopted in the short term.
I hope this analysis helps you see the hidden but strong connection between AI and copper beyond the headlines.