The United States is racing into an energy‑hungry digital era with a power system that was built for a slower, more predictable economy. As artificial intelligence, cloud computing, and electrification surge, Oklo Inc. chief executive Jacob DeWitte is warning that the grid is falling dangerously behind and that the country is courting a capacity crunch unless it builds faster and smarter. His message is blunt: without a step change in investment, planning, and workforce, the next wave of data centers and industrial growth will collide with hard physical limits.
That warning is not arriving in a vacuum. Federal forecasters already see electricity use bending sharply upward after years of relative flatness, while grid planners and energy analysts are flagging the same collision between soaring demand and sluggish infrastructure. The question is no longer whether the United States needs more power, but whether it can deliver it in time and in the right places.
Demand curve jolts awake in the AI and data era
For most of the past decade, U.S. electricity demand barely budged, even as the economy grew, thanks to efficiency gains and a shift toward services. That pattern is now breaking. Official projections show consumption rising again as factories re‑shore, vehicles plug in, and digital infrastructure multiplies, with the latest outlook pointing to a clear upturn in power use after years of stagnation. I see that inflection as the backdrop for DeWitte’s alarm: the grid is being asked to do more, faster, than its planners expected even a few years ago.
The most aggressive pressure point is the data economy. Analysts tracking the buildout of server farms report that Data centers proliferating across the country will require 22% more grid power by the end of 2025 than they did one year earlier and are on track to nearly triple their draw by 2030. Separate research finds that U.S. Power Demand Hits by Data Centers, AI, and Grid Constraints, with electricity consumption in the United States already reaching record levels as these facilities cluster near major metros. In that context, DeWitte’s warning is less a hypothetical and more a description of a trend already in motion.
Oklo’s CEO sounds the alarm on capacity and timing
Jacob DeWitte has emerged as one of the more outspoken voices arguing that the United States is not building power capacity fast enough to match this digital surge. In recent comments, the Oklo Inc. chief executive framed the situation as a looming shortfall, saying the grid cannot keep up with the demands of the digital and AI age without a wave of new generation and transmission. His argument is that the country is entering a period where power is as strategic as bandwidth, and that the current pace of approvals and construction leaves the system exposed to a US power crunch scenario rather than a smooth expansion.
DeWitte’s warning is not just about megawatts, it is about geography and timing. He has pointed to specific projects, including a planned installation at the Pike site, as examples of how new nuclear and advanced generation could be co‑located with large loads to relieve stress on regional grids. In his view, the United States needs to build out both the power grid and new installations like Pike in parallel, or risk forcing data‑center operators and manufacturers into a scramble for scarce capacity. When he says “we are going to need a lot more,” he is effectively arguing that the current project pipeline is mismatched to the scale of the digital buildout already underway.
Data centers, Meta, and the scramble for dedicated supply
Nowhere is the mismatch between demand and infrastructure clearer than in the data‑center sector. Companies racing to deploy AI and cloud services are locking in power years in advance, often in regions where the grid is already constrained. One analysis notes that a once‑steady U.S. electricity demand curve has “jolted awake” as Changes to the Electric Grid and How to Prepare are reshaping planning, with Data centers rapidly changing the supply mix and forcing utilities to rethink where and how they add capacity. I read that as a sign that the old assumption of flat demand is no longer safe, especially in fast‑growing tech corridors.
Oklo Inc. is already positioning itself inside this scramble. Earlier this month, the company’s stock came under pressure even as it announced an agreement aimed at supporting Meta power needs for its data centers, with pre‑construction activities expected to begin in 2026 for a dedicated facility. That deal underscores how large technology firms are no longer content to rely solely on the shared grid; they are seeking bespoke generation to guarantee supply and decarbonization. DeWitte’s broader message, that the grid is falling behind, is reinforced by the fact that hyperscalers like Meta are effectively building their own lifeboats rather than trusting that traditional planning will deliver enough electrons where they need them.
Labor bottlenecks threaten the buildout
Even if the political will and capital exist to build more plants, DeWitte argues that the United States faces a more prosaic constraint: people. Nuclear company Oklo Inc is warning that shortages of skilled workers are threatening to become a bottleneck in efforts to build new power plants, from welders and electricians to specialized nuclear engineers. In my view, that labor gap turns what might look like a straightforward engineering challenge into a race to train and recruit, especially as other infrastructure projects compete for the same talent.
DeWitte has been explicit that this is not a distant risk but a present constraint on timelines. In a separate discussion of Nuclear company Oklo Inc. and its plans, he highlighted that labor shortages may hinder the buildout, and that without a concerted push to expand the workforce, even fully financed projects will struggle to break ground on schedule. A related summary of Labor Shortages May Plant Buildout, Oklo CEO Says, reinforces that he sees workforce as a gating factor for investment in new energy infrastructure. If that assessment is right, then solving the power crunch will require as much focus on apprenticeships and training pipelines as on permits and financing.
Policy, planning, and the race to catch up
Behind DeWitte’s warnings is a broader debate over how quickly the United States can adapt its grid to a more volatile, data‑driven demand profile. Analysts tracking system changes argue that utilities and regulators must move from incremental upgrades to more anticipatory planning, especially in regions where AI and industrial projects are clustering. One recent Report on Changes to the US Electric Grid and How to Prepare stresses that Data centers are rapidly changing the supply mix and that grid operators need new tools to forecast and respond. I see that as a call for closer coordination between tech companies, utilities, and policymakers so that capacity is built ahead of, not behind, demand.
At the same time, federal energy projections are starting to reflect this new reality, with the short‑term outlook now embedding higher growth assumptions tied to industrial policy, electrification, and digital infrastructure. DeWitte’s argument, echoed in coverage that notes we are On Opposite Sides Of The Aisle, But We Kn share concern about grid adequacy, is that this recognition has not yet translated into the scale or speed of concrete projects needed to avoid a crunch. As I weigh the evidence, the picture that emerges is not one of inevitable crisis, but of a narrowing window: if the United States can align investment, labor, and planning in the next few years, it can turn the AI power surge into an engine of cleaner growth. If it cannot, the warning from Oklo’s CEO may look less like a forecast and more like an after‑action report.
More from Morning Overview