Low Energy Fridays: AI to the Energy Rescue

America has always been a nation of builders. From the Model T to SpaceX, and from the Empire State Building to the iPhone, Americans have repeatedly shown an aptitude for developing innovative technologies and building them rapidly and at a high level of quality. Americans’ continued ability to do this has been key to our success. 

The latest technological challenge America has taken on involves artificial intelligence (AI). AI promises to spur advances in everything from culture to healthcare. Running AI, however, requires more than just technical creativity and know-how; it requires a huge amount of electricity. AI data centers consume an enormous amount of power. One large data center alone can use more electricity than a medium-sized city. 

The projected growth in demand for electricity from AI is so substantial that it has fundamentally reordered the nature of electricity planning. For decades, electricity consumption in America was flat or declining for a number of reasons. Many of our devices became more energy efficient, while energy-intensive industries moved offshore. Now, however, the U.S. electricity market is undergoing a fundamental change. Electric demand is projected to increase by 80 gigawatts by 2030. 

The projected increase in electricity demand isn’t, however, all due to data centers. Other sources, from electric vehicles to reindustrialization, also play a role. But regardless of source, America will have to build a lot of new generation to meet all this demand. 

America used to build things much faster. In recent decades, however, an increase in rules and regulations involving project permitting has slowed our ability to build infrastructure to a snail’s pace. At the beginning of the 20th century, the first 28 stations of the New York City subway were built in less than five years. By contrast, a three-station expansion of the subway system begun at the turn of the 21st century took 17 years to complete. 

It’s the same story when it comes to energy. Transmission lines take an average of seven to 10 years to be built. The bulk of this time comes not from the actual construction but from permitting timelines. Building power plants can also be an ordeal, with supply chain issues and permitting requirements delaying projects for years on end. 

If America is going to continue to be a technological powerhouse, it will have to solve its permitting problem. And ironically, AI may not only be one of the causes of our electricity demand growth spurt, but also a critical part of the solution. The U.S. government’s Idaho National Laboratory recently announced that it is partnering with Microsoft to use AI to speed up the permitting process for nuclear power plants. And the Trump administration has set a goal of reducing the time required for permitting nuclear plants from the current decade-long affair that it is to a mere 18 months. AI programs can help with this process by helping to generate application materials and compile safety data necessary for project approvals much more quickly.

Digital automation can help with both the application and approval processes. The Department of Energy has developed an app that local governments can use to quickly review and approve rooftop solar projects. Similar software might soon be used to help regulators resolve applications at record speed.

AI and other computer technologies are not a panacea for permitting delays. Our current system is too cumbersome and creates too many avenues for small numbers of anti-growth ideologues to block needed projects. Still, AI is a way to do permitting quicker and better, which shows that America still has the wherewithal to achieve great things, if only we have the will to do so.