Study Finds AI Energy Consumption Will Surpass Bitcoin's By Year's End
Do you remember how Elon Musk crashed the Bitcoin price when he tweeted that Tesla would no longer accept it due to environmental concerns, and how everyone started worrying about the environmental impact of Proof-of-Work mining? That happened in 2021, and many people still remember it.
But today, Musk's xAI is building what may be the world's largest AI supercluster, and governments are rushing to pass laws to support AI innovation, with few asking about its energy consumption.
A new peer-reviewed study published in the scientific journal Joule suggests that artificial intelligence could account for up to 49% of global data center electricity consumption by the end of 2025, surpassing even Bitcoin's known energy appetite.
Alex de Vries-Gao, a PhD candidate at Vrije Universiteit Amsterdam and a critic of Bitcoin’s energy consumption, found that AI energy consumption could reach 23 gigawatts by January 1, which equates to about 201 terawatt-hours per year. Bitcoin currently consumes about 176 TWh annually.
Image: Joule
“Big tech companies are aware of this trend, with giants like Google even indicating that they are facing a ‘capacity crisis’ in their efforts to expand data centers,” de Vries-Gao noted on LinkedIn. “At the same time, these companies prefer not to share specific numbers.”
“We haven’t seen anything like this since ChatGPT created all the buzz around AI,” he added. “As a result, it remains nearly impossible to get a clear picture of the actual energy consumption of AI.”
Unlike Bitcoin’s transparent energy consumption, which can be calculated from the network’s hashrate, AI’s energy needs remain purposefully opaque. Companies like Microsoft and Google have reported increases in energy consumption and carbon emissions in their 2024 environmental reports, citing AI as a key driver of this growth. However, these companies only provide aggregated data for their data centers, without specifically singling out AI consumption.
Since tech giants don’t disclose AI-related power consumption data, de Vries-Gao started keeping an eye on chips. He looked closely at Taiwan Semiconductor Manufacturing Company’s chip packaging capabilities, since virtually every advanced AI chip requires unique technology.
The logic, de Vries-Gao explained, works like the analogy of a business card. If you know how many cards fit on a sheet of paper and how many sheets a printing press can handle, you can figure out the total production. De Vries-Gao applied this logic to semiconductors, analyzing earnings reports in which TSMC executives acknowledged “very limited capacity” and an inability to “fulfill 100% of customer requests.”
His conclusion: Nvidia alone was estimated to use 44% and 48% of TSMC's CoWoS capacity in 2023 and 2024, respectively. If AMD took another share, the two companies could produce enough AI chips to consume 3.8 GW of power before even considering other manufacturers.
Image: Joule
De Vries-Gao forecast that AI consumption will reach 23 GW by the end of 2025, barring additional production growth. TSMC has already confirmed plans to double its CoWoS capacity in 2025.
Demand for electricity is unlikely to slow. Nvidia and AMD have announced record revenues, and OpenAI has unveiled Stargate, a $500 billion data center project. Indeed, AI is the most profitable business in the tech sector, with each of the world’s top three tech companies surpassing the total market cap of the entire crypto ecosystem, which is $3.4 trillion.
So the environment will probably have to wait.
Source: cryptonews.net
