How AI is driving the boom in data centers and energy demand

WWhile AI could change the world in many unforeseen ways, it is already having a huge impact: a voracious consumption of energy. Generative AI does not simply float on ephemeral intuition. Instead, it gains power through thousands of computers in data centers around the world, which are constantly operating at full speed. In January, the International Energy Agency (IEA) forecast that global electricity demand from data centers will more than double between 2022 and 2026, with AI playing an important role in this increase.

AI industry insiders say the world has enough energy capacity to absorb this increase in demand and that improvements in technological efficiency could offset these increases. But others are skeptical of the industry’s promises and fear that its aggressive consumption of fossil fuel energy sources could harm collective efforts to combat climate change.

See how we got to this point and what the rise of data centers could mean for communities locally and globally.

What are data centers and why are they expanding so quickly?

When someone’s phone data is backed up to the “cloud,” it is actually stored in data centers: massive facilities filled with thousands of constantly running computer servers. In the era of 5G and cloud-based storage, data centers have become essential infrastructure cogs, supporting everything from financial transactions to social media and government operations. Data centers need a continuous and stable supply of energy to function. They now represent more than 1% of global electricity consumption, according to for the AIE.

Data centers were already increasing enormously in number before AI. The bitcoin mining industry played a role in this increase: report from the Energy Information Administration found that bitcoin mining accounted for 2% of the country’s total electricity demand in 2023.

But the technology industry’s strong shift toward AI has increased its construction and use even more dramatically. This is because training AI models is energy-intensive, which consumes energy at a much higher rate than traditional data center activities. A ChatGPT query, for example, uses ten times more energy than a standard Google query, says David Porter, vice president of the Electric Power Research Institute. Porter says that while 10-20% of US data center energy is currently consumed by AI, that percentage will likely “increase significantly” in the future.

This energy usage has been exacerbated by fierce competition among leading technology companies as they race to build more powerful generative AI models. Researchers recently discovered that the cost of computing power needed to train these models is doubling every nine months, with no slowdown in sight. As a result, the IEA predicts that within two years, data centers could consume the same amount of energy as Sweden or Germany. Similarly, researchers at UC Riverside estimated that global demand for AI could cause data centers to consume over 1 trillion gallons of freshwater by 2027.

Despite these staggering numbers, the exact power consumption of many AI models remains opaque. “For a long time, companies like Google and Meta were pretty transparent: until the launch of ChaptGPT, honestly,” says Sasha Luccioni, AI researcher and climate lead at AI platform Hugging Face. “In the last year and a half, they have become much more reserved regarding data sources, training time, hardware and power.”

See more information: The Billion Dollar Price of Building AI

Threatening climate goals and energy infrastructure

This rapid increase in energy use threatens to derail the climate promises that major technology companies set for themselves before the AI ​​hype cycle. In 2020, for example, Google Set a goal run on carbon-free energy 24/7 by 2030. Microsoft made a similar pledge the same year, pledging to become carbon negative within a decade.

But last year, Microsoft took a huge step back in this regard, increasing its greenhouse gas emissions. by 30%, mainly due to its ambitious AI activities. “In 2020, we unveiled what we call our carbon moonshot. This was before the explosion of artificial intelligence,” said Brad Smith, president of Microsoft. told Bloomberg. “So in many ways the Moon is five times further away than it was in 2020, if we just think about our own prediction for the expansion of AI and its electrical needs.”

Microsoft, which invested billions in OpenAI, spent more than US$10 billion in recent quarters in cloud computing capacity and is planning double your data center capacity. In Goodyear, Arizona, which faces water shortages, Microsoft data centers expected to consume more than 50 million gallons of drinking water every year.

Some locations are resisting the construction of data centers. There is currently a de facto situation moratorium against them in Dublin, as they already consume almost a fifth of Ireland’s electricity. But data centers are growing massively elsewhere — including Northern Virginia, colloquially known as Data Center Alley. On the outskirts of Washington DC, historically residential tracts of land are being quickly rezoned as industrial to make way for data centers, drawing the ire of local citizens. Because it is cheaper for companies to build data centers in locations with robust power sources and existing infrastructure, many of them cluster together.

Data centers are now the “number one issue we hear from our voters,” says Ian Lovejoy, a Republican state delegate in Virginia. In addition to quality-of-life concerns, he says local politicians and residents are concerned about data centers threatening access to electricity and water, as well as the idea that taxpayers may have to foot the bill for utilities. future electrical lines.

“Currently, there is not enough power generation or transmission capacity to power the data centers that are under development,” says Lovejoy. “Everyone is betting that we will build the infrastructure and everything will be fine. But if data centers outpace power generation, you could see power outages and a real power constraint situation increasing.”

Of the 8,000 data centers that exist globally, about one third are in the US, compared to 16% in Europe and almost 10% in China. Hong Kong-based think tank China Water Risk estimates that data centers in China consume 1.3 billion cubic meters of water per year – almost double the volume that the city of Tianjin, where 13.7 million people live, uses for homes and services.

Potential solutions

The companies that consume all this energy say they are working on solutions. Many AI Boosters argue that their technology will be crucial to combating climate change, in part by combating inefficiencies. In 2016, for example, Google announced that its DeepMind AI helped reduce its company’s data center cooling energy usage by “up to 40 percent.”

Improvements in chip hardware efficiency can also have a big impact on reducing power usage. This year, NVIDIA released a new line of GPUs (graphics processing units) with energy consumption 25 times lower than its previous models. However, Luccioni believes that any hardware efficiency gains could be offset by Jevons paradox, named after a 19th-century British economist who noticed that as steam engines became more efficient, Britain’s appetite for coal increased. “The more a resource becomes more efficient, the more people will use it,” she says.

Industry insiders also argue that they are leading the way in the energy transition by encouraging the construction of renewable energy sources, particularly in wind, solar and nuclear energy. But many new data centers are still powered by fossil fuels, especially since it takes some time for new energy sources to come online. “Unfortunately, many renewables do not meet the needs of high-speed computing in their demand for consistent, high-quality power,” says Paul Prager, CEO of bitcoin mining company Terawulf, which is also expanding into AI. . “And you can’t snap your fingers and have a plant tomorrow. I think we’re going to have a short-term period over the next three to four years where we see energy costs increase a little bit at the local level.”

Regulation may be on the way. Singapore announced a sustainability standard for data centers in tropical countries last year, and the European Commission moved to regulating the sustainability of data centers across the EU. In the US, Massachusetts Senator Ed Markey presented a bill to study the environmental impacts of AI in February, and the House Energy and Commerce Committee held a hearing on AI energy usage in June.

“Transparency is currently our main obstacle, because many people do not realize the environmental impacts of AI: and not just consumers, but also policymakers and companies,” says Luccioni. “Once we establish transparency and start to have a little more information, we can start trying to regulate.”