The rapid advancement of artificial intelligence (AI) is fueling unprecedented growth in computational demand—and with it, a staggering rise in energy consumption. According to a new analysis, AI's power usage is on track to exceed that of global Bitcoin mining by the end of 2025, marking a pivotal moment in the digital economy’s environmental footprint.
Currently, AI already accounts for up to 20% of data center energy consumption, a figure expected to grow significantly in the coming years. If trends continue unchecked, AI could soon consume nearly half of all power used by data centers worldwide—surpassing the energy-intensive process of cryptocurrency mining, once considered the most power-hungry sector in tech.
The Growing Energy Footprint of Artificial Intelligence
At the forefront of this research is Alex de Vries-Gao, a PhD candidate at the Institute for Environmental Studies at Vrije Universiteit Amsterdam. Known for his work tracking the environmental impact of cryptocurrencies through his platform Digiconomist, de Vries-Gao has now turned his focus to AI’s escalating energy demands.
While exact figures remain elusive—largely due to limited transparency from major tech companies—de Vries-Gao used supply chain data on AI-specific hardware, such as GPUs and TPUs, to estimate current and future energy use. His findings reveal a troubling trend: even as chip efficiency improves, the sheer scale of AI deployment is driving energy demand upward at an accelerating pace.
"Even with more efficient technology, we're seeing exponential growth in demand," de Vries-Gao noted. "The industry is scaling so fast that gains in efficiency are being outpaced by volume."
This surge is primarily driven by the training and operation of large language models (LLMs) like those powering ChatGPT, Gemini, and other generative AI tools. These models require massive computational resources, often running for days or weeks across thousands of specialized processors.
👉 Discover how next-gen technologies are shaping global energy trends.
Why AI Is Becoming a Power-Hungry Giant
Several factors contribute to AI’s growing energy appetite:
- Model Size Expansion: Modern AI models now contain hundreds of billions of parameters. Training them requires repeated calculations across vast datasets, consuming enormous amounts of electricity.
- Inference Workloads: Beyond training, running AI models for everyday tasks—like answering queries or generating images—also demands continuous power.
- Data Center Proliferation: Cloud providers are rapidly expanding infrastructure to support AI services, leading to higher baseline energy use.
- Competitive Pressure: Companies race to launch bigger, faster models, often prioritizing performance over efficiency.
De Vries-Gao initially believed his research into digital technology’s environmental costs might wind down after proof-of-stake transitions in cryptocurrencies like Ethereum drastically reduced their energy use. But the emergence of generative AI changed everything.
“I thought, ‘Oh no, here we go again,’” he said. “It’s another high-consumption technology emerging in a competitive market where sustainability isn’t the top priority.”
This renewed concern underscores a broader issue: technological innovation often outpaces environmental accountability.
Comparing AI and Bitcoin Mining Energy Use
For years, Bitcoin mining dominated conversations about digital energy waste. At its peak, the Bitcoin network consumed over 150 terawatt-hours (TWh) per year—more than many countries.
However, as renewable energy adoption increased in mining operations and network efficiency improved slightly, growth in Bitcoin’s power demand stabilized. Meanwhile, AI’s energy use has entered a steep upward curve.
Estimates suggest that by 2025:
- AI could consume between 100–200 TWh annually, depending on adoption rates and model complexity.
- Data centers overall may use 1,000 TWh or more, with AI accounting for up to 40–50% of that total.
- In contrast, Bitcoin mining is projected to remain around 160–180 TWh, making AI potentially equal or greater in energy draw.
This shift signals a turning point: AI is becoming the new frontier of digital energy consumption.
Environmental Implications and Sustainability Challenges
The environmental implications are profound. Increased electricity demand translates into higher carbon emissions—especially in regions where power grids rely heavily on fossil fuels.
Moreover, the water footprint of AI is often overlooked. Cooling systems in data centers require millions of gallons of water annually. A single large-scale AI model can indirectly contribute to significant water stress in certain geographies.
Sustainability experts warn that without intervention, AI could undermine climate goals. While some companies are investing in green data centers and carbon offset programs, these efforts remain inconsistent across the industry.
👉 Explore how sustainable tech innovations are redefining energy efficiency standards.
Frequently Asked Questions (FAQ)
Will AI really use more energy than Bitcoin by 2025?
Yes, according to current projections. While Bitcoin mining has plateaued in energy growth, AI’s demand is rising exponentially due to widespread deployment and increasing model complexity. By late 2025, AI could match or exceed Bitcoin’s annual power consumption.
Why is AI so energy-intensive?
AI consumes significant power during both training and inference phases. Training involves processing vast datasets across thousands of high-performance chips, while inference requires real-time computation for user interactions—each query adds to cumulative energy use.
Can improved efficiency solve the problem?
Partial gains are possible through better chip design and algorithms. However, efficiency improvements are often offset by increased usage and larger models. This phenomenon, known as Jevons Paradox, means lower costs per operation can lead to higher overall consumption.
Are companies doing enough to reduce AI’s environmental impact?
Some leading firms are adopting renewable energy and exploring liquid cooling and modular data centers. However, transparency remains limited. Without standardized reporting and regulatory pressure, meaningful industry-wide change may be slow.
What role does public awareness play?
Growing scrutiny can drive accountability. Just as concerns over crypto mining influenced policy and innovation, similar attention on AI could accelerate adoption of greener practices and influence consumer choices.
The Path Forward: Balancing Innovation and Responsibility
As AI reshapes industries—from healthcare to entertainment—the need for responsible development has never been more urgent. Policymakers, technologists, and consumers must collaborate to ensure progress doesn’t come at the planet’s expense.
Potential solutions include:
- Mandating energy usage disclosures from tech companies.
- Incentivizing low-carbon AI development through tax breaks or certifications.
- Promoting research into lightweight models and energy-aware algorithms.
- Encouraging edge computing to reduce reliance on centralized data centers.
Ultimately, the goal should not be to halt innovation but to guide it sustainably. As de Vries-Gao’s research highlights, history tends to repeat itself when unchecked technological growth meets limited oversight.
The rise of AI presents both an opportunity and a warning: we have the chance to build a smarter future—one that’s not only intelligent but also environmentally responsible.
By integrating sustainability into the core of AI development, we can avoid repeating the excesses of the past and create a digital ecosystem that serves both humanity and the planet.