Since the introduction of ChatGPT in late 2022, AI has been greeted as a transformational game changer for businesses. However, nearly two years later, organizations’ use of AI is uneven. While some companies have successfully embraced AI, most are taking baby steps with it or still figuring out how to use it in productive, ethical, and secure ways.
One industry sector where AI is already creating a profound impact is data centers.
AI is already transforming data centers like no other technology has in decades. However, this AI-driven data center transformation is also negatively impacting the environment due to AI’s thirst for energy and a corresponding need to cool down the equipment-generated heat created inside a data center.
Without further ado, here are six insights about how AI is transforming data centers and, consequently, the earth.
Data center spending will double this year
As Alphabet, Amazon, Microsoft, and numerous other technology companies seek to become AI heavyweights, they are aggressively expanding their data center footprint.
The most apparent evidence of this data center expansion can be measured in dollar signs. Or Euros.
Last year, the global spend on data center systems grew by 4%. This year, Gartner expects spending to more than double—and reach 10% growth. Gartner says companies will spend $259,680,000 on data centers this year.
The main factor driving this massive jump in data center spending is companies’ desire to incorporate AI capabilities into their products and services.
AI servers are in high demand
Data centers need AI-optimized servers to support AI’s thirst for computational power. And companies—especially hyperscale data centers—want lots of them.
As a result, AI-optimized servers will comprise 67% of hyperscale data centers’ total server spending in 2024, according to Gartner.
General-purpose x86 servers win a second lease on life
To offset their spending on AI-optimized servers, which are expected to peak at $32,300 per server this year, companies are hanging onto their general-purpose x86 servers and replacing them less frequently.
According to Omdia research, the average lifespan of general-purpose x86 servers in enterprise data centers or colocation data centers has increased to 7.6 years. Meanwhile, hyperscale service providers, which aggressively refresh their server fleets, stretched the average life of their x86 servers to 6.6 years in 2023.
Last year, the typical lifespan of a server was three to five years, according to Network World.
AI-driven data centers are very energy-intensive
Because of AI’s craving for computational power and the consequent need to cool data centers’ overheated interiors, the world’s data centers are expected to consume as much electricity as the entire country of India by 2030, according to ARM Chief Executive Officer Rene Hass.
This is particularly startling, given that India is the world’s largest country by population.
In the U.S., AI-driven data centers could consume up to 25% of the nation’s power requirements by the end of this decade, Hass noted.
The bad news: Data centers increase the demand for coal power
Data centers’ AI-driven need for electricity is already adversely affecting utilities’ plans to be greener.
In Northern Virginia’s “Data Center Valley,” some of the largest data centers now use as much electricity as the city of Seattle. The demand for electricity in the area’s hyperscale data centers is causing utilities in nearby Georgia and North Carolina to add fossil-fuel power or consider delaying the shutdown of coal-powered plants, according to The Wall Street Journal.
Reducing coal power has been one of the U.S. energy industries’ recent success stories. For a decade, about 10 gigawatts of coal power had been retired each year. Now, however, that figure will decrease to about 6 gigawatts per year through 2030 because of the big data hub’s increased demand for electricity, according to S&P Global Commodity Insights.
The good news: EU announces new energy efficiency rules
The European Union Commission recently unveiled legislation requiring data center operators in the EU to report their key performance indicators, such as water and energy usage. The regulations aim to rate the sustainability of EU data centers and eventually reduce energy and water consumption. A related goal is to increase data centers’ use of renewable energy, promote grid efficiency, and more.
Experts expect many countries will adopt the same or similar reporting requirements.
Meanwhile, in the U.S., Sen. Ed Markey introduced The Artificial Intelligence Environmental Impacts Act of 2024, which aims to develop standards to measure and report AI’s environmental effects.
“The development of the next generation of AI tools cannot come at the expense of the health of our planet,” he said when introducing the bill.
The bill has yet to pass either the Senate or the House of Representatives.
