AI May Be Ruined Data Centers

Two headlines described what could stop AI expansion in its tracks. The New York Times reported, “Wall Street Is Growing Skeptical of the Data Center Boom.” Bloomberg wrote, “New Data Centers Worth $68 Billion Disrupted in US, Data Show.” (The information came from Data Center Watch and included an analysis of four data centers.) The centers had been blocked or delayed between April and June.

Another widely accepted count of how data centers are performing comes from “The Information.” It puts data center bans at over 500.

Data centers are the power centers of AI. Without them, one could say there is no AI at all. It is an engine without tires, or a plane without wings. If AI is supposed to be the greatest invention in history, it could be crippled by something as old as brick-and-mortar.

Data center investment for 2030 has been put as high as $1 trillion. Some put the investment level at $650 billion this year. Major public companies, including Microsoft (NASDAQ: MSFT) and Alphabet (NASDAQ: GOOG), are leading this expansion. Private companies like OpenAI and Anthropic are joining them, and they expect to use the tens of billions raised by their IPOs to expand AI infrastructure. As these tech companies drain their balance sheets, Wall St.’s largest firms have begun pouring even more money into the sector.

The Times points out that several companies that planned to go public using their data center building plans won’t go public soon. This raises the question of whether they will run low on money.

A new Gallup poll sums up the AI data center challenge. Titled “Americans Oppose AI Data Centers in Their Area,” it reported, “Seven in 10 Americans oppose constructing data centers for artificial intelligence in their local area, including nearly half, 48%, who are strongly opposed.”
Candidates across the political spectrum are running on platforms to stop the buildout of data centers in their districts, cities, and states. It is one of the few issues that unites both Democrats and Republicans.

In the short term, no solution appears to exist. Obviously, tech companies and data center builders have looked almost everywhere in the country where there is room for centers, which can be as large as cities and use up to five million gallons of water a day.

An even larger concern than water is the use of electricity. A large data center can use as much electricity as all the homes in a large city. As these centers compete for the electricity used by homes and legacy businesses, worries about them extend from jumps in electricity rates to blackouts. Part of America’s grid built decades ago may not be able to handle the load. And the demand for electricity is so great that the use of fossil fuels may well rise.

The one problem AI has not solved is how to build data centers to continue its success.

Contact [email protected] for any questions or corrections.