AI Bottleneck Shifts from Chips to Electricity as Natural Gas Powers Data Centers
The bottleneck in artificial intelligence development has shifted from semiconductor chips to electricity. According to Dylan Jovine, founder of Behind the Markets, companies are now struggling to secure power for their data centers and hyperscalers.
Nuclear and geothermal energy dominate the headlines, while natural gas is behind-the-scenes powering midstream operators like Williams Companies (WMB), Enterprise Products Partners (EPD), and Energy Transfer (ET). These companies are building new pipelines, signing long-term supply deals, and constructing power plants to meet the growing demand for electricity.
The AI boom has created opportunities across various industries, but the primary constraint is no longer compute power. Jensen Huang, NVIDIA's CEO, has emphasized that energy availability is the limiting factor, not processing power. Gas turbines are being used as a reliable source of energy, and their sizeable construction costs are being offset by long-term contracts.