AI Has a Power Problem: The Economics of the Data Center Boom
Electricity, grid capacity, land, chips and capital turn AI into an infrastructure problem.
Topic guide
AI Economics
Artificial intelligence is not only a technology story. It is increasingly an investment, infrastructure, productivity, and macroeconomics story.
Why AI economics is becoming infrastructure economics
Artificial intelligence is often discussed as software, but the investment cycle increasingly depends on physical inputs: data centers, semiconductors, electricity, transmission, cooling, land, and financing. That shift changes the questions economists should ask about the technology.
The most important constraints may therefore appear outside the model itself. Grid interconnection times, power prices, capital costs, and the ability to build at scale can determine where compute is deployed and whether large AI investments earn adequate returns.
Why productivity and inflation can move in opposite directions
AI can expand productive capacity in the long run while creating strong investment demand in the short run. Firms can spend heavily on compute and infrastructure before the economy has realized the corresponding productivity gains.
That sequencing matters for inflation and interest rates. A technology can eventually lower unit costs while its buildout temporarily increases demand for scarce capital, labor, electricity, and equipment.
What If AI Raises Inflation Before It Raises Productivity?
Why AI investment can add demand before productivity gains fully arrive.
This Isn't an AI Bubble. It's a Capital Misallocation Crisis.
Capital allocation, infrastructure returns, and the economics behind the AI-bubble debate.
When Bans Don't Matter: What DeepSeek's Nvidia Workaround Really Reveals About the AI Race
Semiconductor access, export controls, and the strategic economics of AI supply chains.