The last time America seriously built nuclear power plants, the Soviet Union still existed. Now Meta is signing 1.2 GW nuclear campus deals with pre-revenue startups, FERC is rewriting grid connection rules on a June deadline, and uranium spot prices are at their highest level since before the iPhone existed. Something has structurally changed — and it's not just the vibes.
This Week in Voltage: FERC Draws a Line in the Sand
On April 16, FERC issued an order setting a hard June 2026 deadline to rewrite the rules governing how large power users — defined as loads above 20 MW — connect to the interstate transmission grid. The docket, RM26-4-000, is specifically framed around the "unprecedented current and expected growth of large loads such as data centers." FERC has reviewed more than 3,500 pages of public comments and held numerous stakeholder meetings. This is not a preliminary inquiry. This is a commission that has decided the current interconnection framework is broken and is moving to fix it before the AI buildout runs into a wall of regulatory friction.
The December 2025 PJM order — directing the nation's largest grid operator to establish transparent rules recognizing large-load co-location — was the opening shot. The June rulemaking is the doctrine. What FERC is building, in real time, is the legal and procedural infrastructure for a world where a hyperscaler can sign a power purchase agreement with a nuclear plant, co-locate a data center on the same substation, and not spend five years fighting interconnection queue bureaucracy.
This matters enormously for the nuclear renaissance thesis. Advanced reactors are not just power plants — they are baseload anchors for the AI economy. But anchors are useless if you can't connect them to the load. FERC is solving the connection problem. The June deadline is the milestone to watch.
Deep Charge: Why Big Tech Became the Nuclear Industry's Best Friend
For thirty years, the nuclear industry's problem wasn't physics or even public opinion — it was economics. Plants cost too much, took too long, and competed against cheap natural gas in a deregulated market that rewarded marginal cost over reliability. Utilities couldn't justify the capital. Ratepayers couldn't absorb the risk. The result was a construction drought that lasted a generation.
What broke the drought wasn't a policy breakthrough or a safety innovation. It was a customer.
Reuters reported that Big Tech is reshaping nuclear funding by inking deals that offer nuclear companies both capital and a clearer commercial path — with Amazon among those actively working on next-generation nuclear arrangements. The pattern is now unmistakable: hyperscalers need power that is always on, carbon-free, and immune to weather. Nuclear is the only technology that checks all three boxes at scale. So they're not waiting for utilities to build it. They're funding it themselves.
The Oklo-Meta deal crystallizes the logic. Oklo has signed an agreement with Meta to develop a 1.2 GW nuclear power campus dedicated to AI data center support. Meta is directly backing next-generation reactors to secure power for its AI workloads. For Oklo — still pre-revenue, still working through NRC processes — this is transformative. It converts a technology bet into a commercially anchored project with a named customer, defined capacity targets, and the kind of long-duration power agreement that makes project financing tractable.
This is the structural shift that changes everything. The old nuclear financing model required utilities to absorb construction risk on behalf of ratepayers, with regulators approving cost recovery after the fact. The new model has hyperscalers writing checks upfront because reliable baseload is a strategic input to their core business — not a commodity they can substitute away from. When Meta needs 1.2 GW of always-on power for AI inference, it cannot run that workload on solar-plus-storage. The physics don't work. Nuclear is the answer, and Meta knows it.
The Microsoft-Nvidia "AI for nuclear" partnership — aimed at streamlining permitting, design, and operations — extends the logic further. If AI can compress the design review cycle, reduce engineering errors in construction documentation, and optimize reactor operations, the two biggest cost drivers in nuclear (time and complexity) become addressable. That's not a marginal improvement. That's potentially a civilizational unlock.
Meanwhile, the regulatory pipeline is filling. On April 15, First American Nuclear submitted its regulatory engagement plan for its EAGL-1 fast-spectrum SMR to the NRC, formally opening pre-application engagement. The company claims EAGL-1 is designed from the ground up to be licensable under existing NRC regulations — a deliberate design choice that sidesteps the years-long process of establishing new regulatory frameworks for novel reactor concepts. Their "Bridge Power" solution offers gas-fueled interim capacity while the nuclear license clears, which is either a clever commercial hedge or an admission that the timeline is long. Probably both.
The point is: the pipeline is real. TerraPower's construction permit. Oklo's Meta campus. First American Nuclear's NRC engagement. These are not press releases about future intentions. They are regulatory filings, signed commercial agreements, and approved permits. The nuclear renaissance has moved from aspiration to execution.
By the Numbers
$105/lb — Uranium spot price as of late March 2026, an 18% gain since January and the highest level since the pre-Fukushima era. Unlike the 2007 speculative spike, this rally is underpinned by simultaneous capacity expansion across the U.S., Europe, Japan, South Korea, China, and India.
18 — Number of operational reactors in Japan's fleet as of mid-March 2026, up from 9 in early 2025, following the accelerated restart of Onagawa Unit 2, Takahama Units 3 and 4, and Shimane Unit 2. Japan's nuclear share is projected to reach 15% of electricity generation by Q3 2026, per CommodityNode's March 21 analysis.
26 — Chinese reactors currently under construction, with plans to double nuclear capacity to 100 GW by 2035.
$12 billion — DOE Loan Programs Office conditional loan guarantees committed to next-generation nuclear projects, including NuScale's SMR deployment in Idaho and TerraPower's Natrium reactor in Wyoming.
50 billion euros — Estimated government-backed financing unlocked by the European Commission's March 5 classification of nuclear as a "strategic net-zero technology" under the Net-Zero Industry Act, covering reactor life extensions and new builds across France, the Czech Republic, Poland, and Finland.
3,500+ — Pages of public comments FERC reviewed before setting its June 2026 deadline to rewrite large-load interconnection rules.
What We're Fighting For
Here is the civilizational argument, stated plainly: baseload abundance is the prerequisite for everything else.
You cannot run a serious AI economy on intermittent generation. You cannot smelt green steel, synthesize green ammonia, or power the compute clusters that will design the next generation of reactors on power that disappears when the wind stops. The energy transition's dirty secret is that it has been building the wrong kind of electricity — variable, weather-dependent, geographically constrained — while the demand curve has started bending toward always-on, always-available, location-flexible power.
Nuclear is the correction. Not because renewables are bad — they're essential, and we need more of them — but because a civilization that wants to run at full capacity needs a foundation that doesn't depend on meteorology. The Kardashev scale doesn't have a "weather permitting" footnote.
What's happening right now is the convergence of three forces that have never aligned before: a technology sector with the capital and the motivation to fund nuclear directly; a regulatory apparatus that is finally moving at the speed of the problem (FERC's June deadline is not an accident — it's a commission that understands the stakes); and a global uranium market that is pricing in a demand wave that the supply chain cannot yet meet.
That last point is the constraint to watch. Uranium at $105/lb with supply deficits building means the fuel cost for new reactors is rising even as the construction economics improve. The industry needs to solve the full stack — not just reactor design and permitting, but mining, enrichment, and fuel fabrication — or the renaissance stalls at the fuel rod.
Watch June. FERC's large-load interconnection ruling will either clear the path for nuclear-anchored data center campuses or create new friction. Watch the NRC's processing timeline for TerraPower and the emerging queue of SMR applications. Watch whether the Oklo-Meta campus structure becomes a template that Amazon, Google, and Microsoft replicate at scale.
The future is electric. The foundation is nuclear. And for the first time in a generation, the money, the policy, and the technology are all moving in the same direction at the same time.
That's what we're fighting for.