AI Companies Build Massive Natural Gas Plants to Power Data Centers

Who doesn't enjoy a classic case of FOMO? From the dot-com boom to Web 2.0, virtual reality to blockchain, the tech world has often been driven by the fear of missing out on the next big thing.

The AI bubble, however, is the granddaddy of them all. Its first offspring — the scramble to secure power for data centers — has now spawned a frantic race to lock down natural gas supplies and equipment. If FOMO could reproduce, the AI bubble would already have grandchildren.

Microsoft announced Tuesday it's partnering with Chevron and Engine No. 1 to build a natural gas power plant in West Texas that could eventually generate 5 gigawatts of electricity. This week, Google confirmed it's collaborating with Crusoe on a 933 MW natural gas plant in North Texas. And last week, Meta revealed it's adding seven more natural gas plants to its Hyperion data center in Louisiana, bringing the site's capacity to 7.46 GW — enough to power the entire state of South Dakota.

Anyone else we're missing?

These recent investments are concentrated in the southern U.S., home to some of the world's largest natural gas reserves. The U.S. Geological Survey recently estimated that one region alone holds enough gas to supply the entire country for 10 months. Every data center operator seems eager to claim a piece.

The rush for natural gas has created a shortage of turbines for power plants, with prices expected to rise 195% by the end of this year compared to 2019, according to Wood Mackenzie. Turbines account for 20% to 30% of a plant's cost. The consultancy notes that companies can't place new orders until 2028, and turbine delivery takes six years.

That means tech companies are betting the AI frenzy won't cool down — that AI will keep demanding exponential power, and that natural gas generation is essential for success in the AI era.

They might regret that third assumption.

While U.S. natural gas supplies are plentiful and shipping costs keep the country somewhat insulated from Middle East turmoil, supplies aren't infinite. Recently, production growth in the three major regions — responsible for three-quarters of all U.S. shale gas — has slowed considerably.

It's unclear how insulated tech companies are from price swings, since none have disclosed specific contract terms. Much will depend on how firm those prices are.

Even if contract prices are rock-solid, companies could still face consequences.

Natural gas generates about 40% of U.S. electricity, according to the Energy Information Administration, so electricity prices are tightly linked to gas prices. Tech companies might shield themselves from scrutiny for a while by placing their gas plants behind the meter — connecting directly to data centers instead of the grid. But natural gas isn't unlimited, and if their ambitions grow too large, even behind-the-meter operations could drive up power prices for everyone. We've seen how that plays out.

It won't just be households getting upset. Other industries, especially those more dependent on natural gas and unable to switch to renewables, might push back against data centers hoarding the resource. Powering a data center with wind, solar, and batteries is straightforward. Running a petrochemical plant? Not so much.

Then there's the weather. A harsh winter could shift the calculus by boosting household demand. Wellheads might freeze, cutting supplies dramatically — as happened in Texas in 2021. When gas runs short, suppliers face a choice: keep AI data centers running or let people heat their homes?

By snapping up natural gas supplies and moving behind the meter, tech companies can claim they're "bringing their own power" without straining the electrical grid. But in reality, they're just shifting their load to another grid — the natural gas grid. The AI rush reveals just how physically constrained the digital world remains. Does it make sense to bet big on a finite resource? Tech companies might regret giving in to the FOMO.