A Man in Manassas Just Got a $281 Electric Bill. He's Not Alone.

John Steinbach has lived in his Manassas, Virginia home for nearly 40 years. His electric bill used to run around $100 a month. In January 2026, it came to $281.

"It's just so far beyond any bill I've ever had," he told Consumer Reports.

Nothing changed at his house. What changed was the grid around it.

Data centers now consume more than one in four kilowatt-hours generated in Virginia. That's not a projection. That's today. And Dominion Energy received approval for its first base-rate increase since 1992 — roughly $8.51 more per month for residential customers — driven in large part by infrastructure needed to serve that load.

I'm releasing a restricted intelligence briefing on a "Hidden Inheritance."

This is the result of 20 years of work by the U.S. Extended Continental Shelf Project.

In December, federal filings finally revealed the coordinates of this discovery.

American investors can now position themselves for a $500 trillion resource windfall.

This is made possible through one small public company already holding the key partnerships.

Most Americans have no idea the U.S. just added territory larger than Texas and California combined.

I don't know how long I can keep this briefing online before the "insiders" try to pull it.

See the official coordinates and the ticker here.

See the ticker and the $500T Briefing here >>

An NC State-led study published this week makes the math explicit. Data center demand could push electricity costs up 57% in some regions by 2030. The national average: somewhere between 6% and 29% higher. Researchers from NC State, Carnegie Mellon, Pittsburgh, and Toronto modeled power regions across the lower 48. The findings are stark.

In Northern Virginia, the model shows coal-fired generation actually rebounding to meet demand. That's not a typo. The AI boom is causing a coal comeback in one of the country's most data-center-dense regions.

On March 4, the White House gathered seven tech executives for a signing ceremony. Amazon, Google, Meta, Microsoft, OpenAI, Oracle, and xAI all put their names on the "Ratepayer Protection Pledge" — a commitment issued as a presidential proclamation, promising to pay for their own power and cover all infrastructure upgrades. No external enforcement mechanism. No oversight body with teeth. The pledge relies entirely on voluntary compliance and state-level implementation through utility tariffs.

The watchdog that monitors PJM's markets — the largest grid in the US, serving 67 million people — put it plainly last week: "The price impacts on customers have been very large and are not reversible."

John Steinbach's $281 bill isn't a glitch. It's a preview.

Utility shutoffs in Pennsylvania jumped 21% last year. Not because people stopped paying their bills. Because the bills got much harder to pay.

"We are already paying in the neighborhood of a billion dollars more this year in generation costs as a result of that increase in capacity, which is directly attributed to the load growth from data centers," said Elizabeth Marx, executive director of the Pennsylvania Utility Law Project.

On April 30, the Pennsylvania Public Utility Commission took a step most state regulators haven't. It voted to advance a model tariff governing how large loads — data centers drawing more than 50 megawatts — connect to the grid. The guidance isn't binding. But it does something important: it shifts the baseline assumption. Large customers pay for their own infrastructure. The cost doesn't get spread to households.

Regulatory Watch

Pennsylvania's PUC guidance includes two new amendments: large-load customers must build their own infrastructure upgrades where feasible, and utilities must recover all transmission and distribution costs from the data center — not from existing ratepayers. North Carolina lawmakers introduced similar legislation this month, the Ratepayer and Resource Protection Act, requiring data centers with peak demand above 40 MW to pay cost-based rates and generate at least 25% of their power on-site using clean energy.

Next up: the June 2026 PJM capacity auction, which will set wholesale electricity prices for the 2028/29 delivery year. The price cap framework that capped the last two auctions does not extend to this one. Analysts are watching closely.

Thirty states have introduced more than 300 bills on data center policy in 2026. That's not a lobbying effort. That's a voter pressure campaign translated into legislation.

The Ratepayer Protection Pledge that seven companies signed at the White House in March is a starting point, not a finish line. It requires companies to negotiate separate rate structures with utilities and state governments — but that process plays out region by region, regulator by regulator. PJM's market monitor has already said the price impacts are irreversible without faster federal action.

The policy window is open. Whether regulators move fast enough to matter for this summer's auction is a different question.

The AI Buildout Has a New Bottleneck. It's Not Money.

Five years ago, the constraint on a data center project was land. Then it was chips. Now it's electrons — and that's a much harder problem to solve.

The five largest US tech companies are on track to spend over $600 billion in capital expenditure this year. That's up roughly 50% from last year, and more than double the capex intensity seen at the peak of the 1990s internet buildout. The money is there. The grid is not.

5–10 years

Typical wait time for grid interconnection when significant transmission upgrades are required — Berkeley Lab / interconnection queue analysis, 2026

Nearly 2,300 gigawatts of generation and storage capacity are currently stuck in US interconnection queues. That number exceeds the country's entire installed power capacity — sitting in line, waiting for studies, permits, and transformer deliveries that can take years.

PJM projects that 32 gigawatts of new peak demand will hit its grid between 2024 and 2030. All but 2 gigawatts of that comes from data centers. Yet only 2.7 gigawatts of new generation came online in the 12 months before its last capacity auction. The math doesn't work.

Industry analysts at Sightline Climate estimate that close to half of all global data center projects scheduled for completion this year face delays directly tied to power constraints. Morgan Stanley models a 49-gigawatt supply shortfall in the US alone through 2028.

Hyperscalers have started working around the queue entirely. Gas turbines deployed on-site. Dedicated solar farms wired directly to campuses. Behind-the-meter setups that bypass the interconnection process. It's faster. But it also means big industrial customers leaving the shared grid — which, as utility economists have long documented, pushes fixed costs onto the residential customers who remain.

Goldman Sachs flagged the investment risk earlier this year: AI capex growth is slowing relative to the pace set in 2025. Companies spent freely when profits ran two to three times their investment. That ratio is compressing. The firms that can't secure power fast enough won't just fall behind on AI deployment — they'll face a harder story to tell Wall Street.

The constraint on the AI buildout is no longer capital. It's the grid. And the grid is years behind.