Will AI's Power Hunger Eclipse Bitcoin Mining in 2025?

Artificial intelligence (AI) is poised to overtake Bitcoin mining as the world's largest data center energy consumer by the end of 2025, according to a groundbreaking analysis by Alex de Vries-Gao from Vrije Universiteit Amsterdam's Institute for Environmental Studies.

With AI's electricity demands expected to reach 23 gigawatts, equivalent to the entire data center consumption of the United Kingdom, this shift signals a new era of energy challenges for the tech industry.

As global electricity use in data centers is projected to double by 2030, utilities are scrambling to meet demand, sparking a revival of nuclear projects and new gas-fired power plants.

AI's Soaring Energy Appetite

The rapid rise of AI is reshaping the global energy landscape. De Vries-Gao's research estimates that AI could account for nearly half of all data center electricity consumption by 2025, with specialized AI hardware consuming between 46 and 82 terawatt-hours (TWh).

This is a stark contrast to Bitcoin mining, which currently uses around 10 gigawatts. Unlike Bitcoin's steady energy profile driven by specialized ASICs, AI's power needs fluctuate based on training and inference tasks concentrated in large-scale data centers.

Recent data highlights that Nvidia's latest Grace Blackwell chips, while more efficient than predecessors, still require significant power, fueling this unprecedented demand.

The International Energy Agency projects that global electricity use in data centers could hit 945-1,050 TWh by 2030, roughly doubling from today’s levels.

ALSO READ | Nebraska Engineers Unveil Self-Healing Robotic Muscle Mimicking Nature

Regional Impacts and Sustainability Challenges

The United States and China are expected to drive nearly 80% of the global increase in data center energy use by 2030. In the U.S., data centers already consume 2.5% of total electricity (approximately 130 TWh), a figure that could triple to 7.5% (390 TWh) by the end of the decade, enough to power nearly a third of American homes. Europe, by contrast, is projected to see a 70% increase.

This surge is prompting a shift toward renewable energy, with the share of renewables in data center power expected to grow from 27% to 50% by 2030, primarily through expanded wind and solar capacity.

However, AI's concentrated energy demands are pushing utilities to rely on fossil fuels in the short term, with new gas-fired plants under construction and dormant nuclear projects being revived.

Did You Know?
Data centers currently consume as much electricity as the entire global airline industry, accounting for 1-2% of the world’s power usage.

AI vs. Bitcoin: A Tale of Energy Profiles

While Bitcoin mining has faced scrutiny for its energy intensity, AI's rapid growth presents distinct challenges. Bitcoin miners often chase low-cost energy, including renewables, across distributed networks.

AI, however, requires massive, centralized data centers, creating localized strain on power grids. This mirrors historical concerns about cryptocurrency but introduces new complexities due to AI's variable workloads.

The phenomenon exemplifies Jevons' Paradox, where efficiency gains in AI hardware paradoxically drive higher overall energy consumption as adoption skyrockets.

Recent reports indicate that the U.S. is set to launch its first nuclear-powered Bitcoin mining center in early 2025, highlighting the ongoing race to secure sustainable energy for both sectors.