Raj Agrawal, an executive at the global investment firm KKR, has warned that the rapid growth of artificial intelligence could drive electricity demand far beyond current forecasts, potentially straining the world's energy systems. The rising power needs, he said, may ripple through industries that depend on cheap electricity and force a rethink of where money flows in the energy sector.
A warning from an influential investor
Agrawal's comments stand out because of KKR's size and reach. The firm manages hundreds of billions in assets, with deep holdings in energy, infrastructure, and technology. When someone inside that kind of operation flags a trend, the market tends to listen. He didn't mince words: AI's appetite for computing power will translate into an appetite for electricity that many planners haven't fully accounted for. The gap between current projections and what Agrawal sees coming is wide enough to raise hard questions about grid capacity, fuel supply, and carbon targets.
Why cheap power matters more than ever
Data centers that train and run large AI models already consume enormous amounts of electricity. As the technology spreads into more everyday uses — from search to manufacturing to finance — that consumption will only climb. Agrawal's warning suggests the climb could be steeper than most utility forecasts assume. That puts a target on industries like aluminum smelting, chemical production, and steelmaking, all of which rely on low-cost power to stay competitive. If AI eats into that cheap supply, those industries face higher costs or forced efficiency measures. The impact won't be limited to one region; energy markets are global, and a surge in demand in one part of the world can shift prices everywhere.
How it reshapes investment strategies
For investors, the message is clear: the way they evaluate energy assets may need to change. Traditional models that play out slowly over a decade or two might not capture the speed of AI adoption. Agrawal's warning points to a world where power demand grows fast enough to make renewable and gas projects look very different on a balance sheet. That could drive capital toward companies that can quickly add capacity — nuclear operators, grid builders, and firms that specialize in energy-efficient data centers. At the same time, it might pull money away from sectors that assumed low power costs would last forever.
The question hanging over all of this is timing. Nobody knows exactly when AI's energy use will hit its steepest phase. But Agrawal is saying that planners and investors shouldn't wait to find out. If he's right, the window for action is narrower than the industry believes.
