Abundance

by Ezra Klein

AI revolution makes the cause of energy abundance even more urgent. In the last few decades, US energy infrastructure projects have been slowed by all the challenges we’ve described: a lack of productivity in construction, permitting blockages, extended environmental reviews, and long interconnection queues. These bottlenecks are largely self-made, and if we don’t make it easier for AI companies to build in America, we should expect them to build data centers abroad. Some AI executives have met with Gulf states leaders about siting data centers in the Middle East. In the next few decades, trillions of dollars of AI infrastructure could be built somewhere in the world. The biggest question is where. If the US fails to add energy supply in the US, the results could be chaotic, at best, and catastrophic at worst. As new data centers demand more energy, electricity prices would rise for consumers, in the absence of supply growth. More troublingly, it’s conceivable that AI researchers are years, not decades, away from building a superintelligent system with the ability to hack foreign government secrets, cripple their military software systems, and partly collapse the energy grids of adversaries. This would be a breakthrough akin to a kind of digital nuclear bomb.