AI Set to Overtake Bitcoin Mining in Energy Consumption by 2025, Study Finds

Artificial Intelligence (AI) is on track to surpass Bitcoin mining in global electricity consumption, according to a new study forecasting that AI could consume nearly half of all energy used by data centers worldwide by the end of 2025.

The analysis comes from Alex de Vries-Gao, a PhD researcher at the Institute for Environmental Studies at Vrije Universiteit Amsterdam. Known for his prior research on cryptocurrency’s energy impact, De Vries-Gao recently published a new article in the journal Joule exploring the growing energy demands of AI systems.

AI Already Consumes a Fifth of Data Center Power

According to De Vries-Gao, AI currently accounts for up to 20% of electricity used by data centers. This estimate, reported by The Verge and reviewed by Al Arabiya Business, highlights a major shift in the tech sector’s energy landscape.

Despite the growing concern, major tech companies remain largely silent about how much power their AI systems consume, making accurate tracking difficult. As a result, De Vries-Gao turned to hardware production estimates and supply chain data for specialized AI chips to model energy usage.

Rapid Growth Despite Efficiency Gains

Although AI hardware has become more energy efficient, the overall power demand continues to rise due to the massive scale of deployment. By combining industry reports, public hardware specifications, and corporate financial disclosures, De Vries-Gao estimated how many AI devices are in use—and how much electricity they likely consume.

His analysis concluded that AI’s total electricity consumption in 2023 may have rivaled that of the Netherlands. By 2025, this could climb to levels comparable to the United Kingdom, with AI systems potentially requiring up to 23 gigawatts of electricity.

AI Among Key Drivers of Future Energy Demand

These findings align with projections from U.S. consultancy ICF, which recently reported that electricity demand in the U.S. could rise by 25% by 2030. The surge is largely attributed to AI workloads, traditional data centers, and cryptocurrency mining.

However, estimating the precise environmental impact of AI remains complex. Variables such as location, hardware type, and power sources all influence carbon emissions.

For instance, processing AI tasks in data centers powered by coal-heavy grids—like those in West Virginia—can produce double the emissions of centers in more renewable-friendly states like California.

Why Transparency Matters

The intensity of AI’s energy use depends on several factors:

• The size and type of AI models in use

• The complexity and frequency of user queries

• The carbon profile of the local power grid

• Cooling systems and hardware architecture

Because of these complexities, researchers argue that improved transparency from major tech firms is essential. Public disclosure of AI-related energy metrics would allow for more accurate modeling and accountability.

The Efficiency Dilemma

While some companies are investing in more efficient hardware, it remains unclear whether such improvements will be enough to offset the surging demand for AI services.

In a surprising development, Chinese AI firm DeepSeek claimed earlier this year that its models require only a fraction of the electricity consumed by Meta’s "Llama 3.1." This has sparked industry-wide questions: Does AI advancement truly need to be so energy-intensive?

If companies like DeepSeek can prove that cutting-edge AI models can run with drastically lower power requirements, the broader industry may be forced to reevaluate its current energy strategies.

Conclusion: The Future of AI and Energy

As AI continues to evolve, its environmental footprint cannot be ignored. The technology’s rapid growth, paired with limited transparency, makes it difficult to gauge its true impact. What’s clear, however, is that AI is becoming a major contributor to global electricity demand—potentially even outpacing energy-hungry activities like Bitcoin mining.

The challenge now lies in balancing innovation with sustainability. Without significant changes in how AI systems are designed, deployed, and powered, the environmental costs may become as concerning as the technological breakthroughs themselves.