For most of the AI boom, investors have treated the trade like an increasingly expensive trip to the chip aisle. NVIDIA Corp (NASDAQ:NVDA), Advanced Micro Devices Inc (NASDAQ:AMD) and Micron Technology Inc (NASDAQ:MU) became the obvious shorthand for artificial intelligence.
GPUs, high-bandwidth memory and the silicon supply chain kept the machine going.
• GE Vernova stock is showing upward bias. What’s ahead for GEV stock?
But the next leg of the AI economy may be less about what makes the machine bigger and more about what makes it alive. Electricity, power equipment, grid connections, turbines, transformers… the unglamorous yet essential business of getting electrons to data centers on time.
BlackRock Investment Institute is now pointing investors toward that second-order trade.
“The combined pressures of vulnerable energy supply and rising power demand are making energy security a durable investment theme, favoring infrastructure and critical bottlenecks,” the firm wrote in the recent note.
Awakening the Growth
The scale of the domestic reversal in the U.S. is staggering. According to Exponential View’s June 25 report, U.S. electricity net generation went essentially nowhere for 16 years, recording “±0 growth” between 2008 and 2024.
In comparison, the historical average addition between 1950 and 2008 was about 6 terawatt-hours per month, annualized.
Now the line has snapped upward. Since 2024, U.S. electricity generation has been growing by roughly 9 TWh per month, a 50% jump over the old postwar average. A utility sector built for slow-moving forecasts has suddenly been asked to serve a hyperscaler economy running on 12- to 18-month deployment cycles. Still, the problem is not just demand, but the timing.
“Power availability is now a first‑order concern for the data center industry,” ING’s analyst Coco Zhang recently wrote. Traditional grid interconnection for data centers and new power generation, she notes, often takes more than four years. Data centers, meanwhile, need capacity almost immediately.
Jumping the Power Queue
That mismatch is giving rise to a parallel power system.
Instead of waiting politely in the utility queue, hyperscalers and data center operators are planning to generate power directly on-site. ING estimates that more than 55 gigawatts of behind-the-meter capacity is planned for U.S. data centers, exceeding New York state’s total installed capacity. About 75% of that planned U.S. on-site capacity is expected to come from natural gas turbines, engines and fuel cells.
This reality is where the trade widens beyond Nvidia, AMD and Micron.
GE Vernova Inc (NYSE:GEV) is one of the most obvious beneficiaries. Microsoft’s latest 20-year power contract with Chevron for Project Kilby in West Texas shows the model. GE Vernova will provide the majority of the hardware for the 2.67-gigawatt facility.
The company’s Electrification segment has also seen surging demand, including a record $2.4 billion in data center equipment orders in the first quarter of 2026, exceeding its prior full-year total.
Eaton Corporation (NYSE:ETN) is another name sitting directly at the choke point. The firm produces switchgear, smart transformers, power distribution and grid management equipment — precisely the infrastructure data centers need before turning on a single GPU.
Then there is Vistra Corp (NYSE:VST), the independent power producer. For hyperscalers that need dispatchable, continuous power outside the slow-regulated utility system, companies with available generation and the ability to sign long-term bilateral contracts can become very attractive counterparties.
There is also a political catch. A recent Wood Mackenzie’s analysis describes the emerging fight as “a new politics of electricity.”
U.S. residential power prices have risen more than 40% since 2021, and in Ohio — now a major data center hub — retail electricity prices rose 22% year over year. Consumers may not care whether the culprit is AI, weather, fuel, transmission, or old utility regulation. They just see the number on the bill.
For that reason, the AI power trade is not a simple demand story. It is a bottleneck story, a permitting story, a local politics story, and increasingly, an energy security story.
Photo: Shutterstock
Recent Comments