Get your free personalized podcast brief
We scan new podcasts and send you the top 5 insights daily.
An AI query consumes 30 times more electrical energy than a standard Google search. As AI becomes ubiquitous, the resulting demand for electricity—and the copper-intensive infrastructure to generate and transmit it—is projected to grow almost infinitely, creating a structural shortage.
Related Insights
The next major bottleneck for AI, electrification, and defense is not chips, but copper. To meet baseline GDP growth projections—excluding upside from data centers and green energy—the world needs to mine the same amount of copper in the next 18 years as it has in all of human history.
The primary bottleneck for scaling AI over the next decade may be the difficulty of bringing gigawatt-scale power online to support data centers. Smart money is already focused on this challenge, which is more complex than silicon supply.
Even before the AI boom, demand for copper was outstripping supply for standard manufacturing and electrification. The addition of massive data centers and EVs creates a long-term supply deficit that is nearly impossible to solve, as bringing new mines online can take over 15 years.
The massive energy consumption of AI data centers is causing electricity demand to spike for the first time in 70 years, a surge comparable to the widespread adoption of air conditioning. This is forcing tech giants to adopt a "Bring Your Own Power" (BYOP) policy, essentially turning them into energy producers.
To sustain 3% global GDP growth, the world must mine as much copper in the next 18 years as it has in the last 10,000. This excludes the massive additional demand from the energy transition, data centers, and AI, making the supply challenge almost insurmountable.
Daniel Gross's prescient question about copper being mispriced proved correct. The metal hit all-time highs due to AI's physical needs, with a single NVIDIA server rack containing two miles of copper wire. This highlights a critical, non-obvious bottleneck in the AI supply chain.
The rapid expansion of AI is creating an unprecedented surge in electricity demand. Projections show that by 2030, the additional power required will be comparable to the total consumption of Japan, the world's fifth-largest power-consuming nation. This highlights the massive scale of the infrastructure challenge.
For three decades, US power demand was stagnant due to energy efficiency and offshoring. The AI build-out has abruptly ended this era, driving unprecedented ~5% annual growth. This demand shock has created a massive bottleneck in the supply chain for critical hardware, with a new power generation unit ordered today not expected for delivery until 2029.
The primary obstacle to AI's growth is not semiconductor supply but physical power infrastructure. Data centers face a massive power deficit, needing more than double the contracted grid capacity by 2028, with long delays for connections, labor shortages, and local opposition acting as major hurdles.
The rapid expansion promised by AI firms faces real-world bottlenecks. These include shortages of key commodities like copper, insufficient power grid capacity requiring years to build new plants, and a lack of skilled construction labor, making promised timelines highly unrealistic.