The U.S. electricity grid hasn’t faced a demand shock like this since the post-World War II industrial boom. For decades, power consumption grew at a sleepy 1% to 2% annually. Then came data centers. Then came AI. And now, if Nvidia (NASDAQ:NVDA | NVDA Price Prediction) CEO Jensen Huang is right, we haven’t seen anything yet.
So here’s the question investors need to grapple with: if today’s AI already strains the grid, what happens when the technology never sleeps?
The Number That Reframes Everything
At ServiceNow‘s (NYSE:NOW) Knowledge 2026 conference earlier this month, Huang dropped a figure that should be bookmarked by anyone holding energy or tech stocks. He stated that the compute required for agentic AI will rise at least 1,000% compared to generative AI — in just two years.
Generative AI is reactive. You type a prompt, the model burns through tokens, you get an answer, and the GPU cools down. Agentic AI is another animal entirely. Agents read, plan, call tools, write code, query databases, and verify their own work — stringing those steps together for minutes or hours at a time, often without any human in the loop. Each cycle consumes more compute than a dozen chatbot replies.
Huang’s vision is 10 billion digital AI agents working alongside human employees. “The entire manufacturing line will be operated by robots, managed by more robots, and the entire factory is a robot,” he told CNBC’s Jon Fortt.
That is not a modest prediction. And it has a power bill attached.
What It Means for the Grid — and Your Wallet
Let’s start with where we already are, because the baseline is remarkable on its own.
According to a Business Council of Sustainable Energy report, U.S. data centers now draw approximately 41 gigawatts of power — a 150% increase over five years. The Lawrence Berkeley National Laboratory projects that figure rises to between 325 and 580 terawatt-hours (TWh) by 2028, representing up to 12% of total U.S. electricity consumption. The IEA’s April report, Key Questions on Energy and AI, projects global data center electricity consumption doubling from 485 TWh in 2025 to 950 TWh by 2030, with AI-specific data centers tripling their consumption over the same period.
Now layer in Huang’s 10x compute multiplier for agentic workloads, and the math gets uncomfortable fast. Especially since he says he wouldn’t be surprised if that figure is off by “a couple orders of magnitude.”
© 24/7 Wall St.
The grid is already showing cracks. Dominion Energy (NYSE:D), which serves Northern Virginia — the world’s densest concentration of data centers — proposed its first base-rate increase since 1992 in February 2025, and was granted an $11.24 per month for a typical household in 2026. That is not a technology company absorbing costs. That is an ordinary ratepayer subsidizing the AI buildout, and the local backlash is growing nationwide.
In some regions, AI-driven demand is already outpacing available capacity, forcing companies to delay projects or install their own natural gas generators rather than wait years for grid connections.
Granted, efficiency gains are real. Huang himself has argued that Nvidia’s hardware has delivered 100,000x improvement in performance per watt over the past decade. But the IEA’s own data from April shows the counterforce at work: power consumption per AI task is falling, yet total consumption keeps rising, because the number of tasks is growing faster than the efficiency gains can offset. That is the Jevons Paradox in live action.
Where Smart Investors Are Looking
The energy implications of agentic AI are not a future problem. They are a present investment thesis.
The IEA reports that the pipeline of conditional agreements between data center operators and small modular reactor (SMR) nuclear projects grew from 25 gigawatts at end of 2024 to 45 gigawatts by April. Tech companies, with their investment-grade balance sheets and decade-long power contracts, are accelerating nuclear commercialization faster than any government program has managed in 30 years. Google has a power purchase agreement with Kairos Power for SMR capacity by 2030. Amazon Web Services acquired a data center campus adjacent to Talen Energy‘s (NASDAQ:TLN) 2.5-gigawatt Susquehanna nuclear plant in Pennsylvania, securing 1,920 megawatts of dedicated capacity.
Meanwhile, the four largest cloud providers — Amazon (NASDAQ:AMZN), Microsoft (NASDAQ:MSFT), Google, and Meta Platforms (NASDAQ:META) — collectively committed more than $710 billion in AI infrastructure capital expenditures for 2026 alone. The IEA notes that just five technology companies now spend more on capital expenditure than the entire global oil and gas production industry. Renewables are also accelerating, accounting for roughly 40% of all corporate power purchase agreements signed in 2025.
In short, the energy sector is being reshaped by the same forces driving Nvidia’s 65% revenue growth in fiscal 2026 — when the company reported $215.9 billion in annual revenue.
Key Takeaway
Investors focused solely on chip stocks may be looking at the engine while ignoring the fuel supply. The transition from generative to agentic AI is not a software upgrade. It is an infrastructure buildout on the scale of electrification itself — one that touches utilities, nuclear developers, natural gas operators, transmission equipment makers, and the monthly electricity bills of ordinary households.
Huang’s 1,000% compute figure is not a boast about Nvidia’s hardware roadmap. It is a demand signal for every kilowatt-hour between here and 2030. Sharp investors would do well to follow the power lines, not just the silicon.
