Powerstack: FERC’s order puts the power gap on the record

Great to see so many of you this week in London, especially at our event on Tuesday! We'll send out a summary on Monday in our other newsletter, CTVC. Now let's all find a place to hide from this heat.

Read on below for our take on the big FERC news, and more.

This is Currence’s weekly newsletter on the moves and motives shaping the load growth era. Not a client yet?

The Feds are finally taking on large loads. FERC issued "show cause" orders last week to all six RTOs and ISOs under its jurisdiction — PJM, MISO, SPP, CAISO, ISO-NE, and NYISO — directing each to either justify its existing tariff treatment for large loads or file reforms within 60 days. A separate 30-day deadline requires each operator to report how it will ensure adequate generation for existing and incoming large loads. The orders name five reform areas:

Essentially, FERC wants a framework that answers the questions every large load developer is already asking: how do I get grid access, how long does it take, and who pays for the infrastructure I need? The goal is to move large-load interconnection from case-by-case free jazz improvisation toward a tariffed, reviewable standard, which SPP and PJM had already started.

And rather than issuing a Notice of Proposed Rulemaking, which could take years to finalize, FERC used Section 206 of the Federal Power Act to move quickly after a directive from the DOE late last year. This mechanism is a faster tool that signals how urgently the federal government is treating both the pace of large-load growth and the affordability pressure it's creating for existing ratepayers.

Each region will respond differently and their responses (or requests for an abeyance for up to 90 days) will trickle in during the next 60 days, producing six regional answers rather than one national standard. That's partly by design. A uniform federal rule would face stronger legal challenges from state regulators who are watching for signs of federal overreach, and FERC explicitly preserved state authority over generation siting and permitting, retail electricity sales, and construction decisions.

The orders also don't touch the other gates that determine whether a large load project gets built: permitting, fuel supply, equipment, project finance. And there are geographic gaps too. These six operators cover only roughly two-thirds of US electricity load. ERCOT is outside FERC's jurisdiction entirely, so Texas projects stay governed by state permitting and ERCOT coordination. Similarly, the Southeast and the non-CAISO west, where a decent chunk of the current hyperscale build-out is heading, are outside the RTO footprint. 

Now, we wait for the wave of RTO filings due this summer, which will show how each operator intends to respond and where the fights over cost allocation and co-location treatment are likely to land.

The main thing is to keep the main thing, the main thing.

I know I'm a broken record, but to me this is still a speed-to-power or Fastest MW story. I know we’ve posted this chart before, but it holds up.

For sure, the large load reform areas in the order will help. A better understanding of study processes, cost and risk allocation, co-location terms, and flexible-load service terms are much needed. But the cleanest large-load tariff in the country doesn’t put a megawatt online.

So while the 60-day tariff response is important, the filing that’s really got me interested is the 30-day adequacy report – to me, this is the main thing. We’ve clearly seen (and written about) hints of what the RTOs will say, but this will be the first time all six operators have to put the gap between queued load and deliverable generation on the record. Of course, the RTOs are likely to skew the framing a bit, either going optimistic to say ‘we’re open for business’, or going pessimistic to say ‘come build more generation and infra here’.

But either way, the gap will be telling. If they project oversupply or even semi-close to 1:1 for generation supply and large load, it would indicate things will move ahead in the region. But if they come out with figures that show a substantial power undersupply or severely delayed queue, it would follow that large load demand planned for that region would have to explore behind-the-meter or bridging options, another region, or quietly go into delay.

And that supply/demand gap is where the dollars sit. If you know the power supply gap, you can start to estimate the cost of delay for that demand. Back-of-the-envelope, but if we assume roughly $100m a day in infra rent for a 1GW AI factory, a six-month delay means ~$18bn – total that up for a market and that’s some serious cheddar.

While the tariff reform is wildly helpful, the Fastest MW still wins. So let’s mark our calendars for 18 July to watch for those adequacy reports.

** Currence scores individual projects on eight success factors, including or especially their power set-up – Currence clients, see our Q2 Data Center Outlook for full methodology and analysis.

Anyone with a megawatt. Grid-scale generators in the named RTOs are the obvious winners, but the order's explicit focus on nearby generation opens a lane for anyone supplying power near a large load, not just co-located on the same campus.

Behind-the-meter and DER suppliers. The orders put co-location and BTM treatment on the table for all six regions, not just PJM, where FERC already ruled its co-location and BTM rules were unjust in December 2025 and forced a redo. This could be a boost for companies like Bloom Energy, which has been pitching fuel cells as grid-independent solutions for AI factories, and aggregators like Voltus or Uplight that can package flexible demand response.

Flexible large loads. The barrier has always been the lack of a pathway to trade firm for fast, and flexibility is now finally on FERC’s radar. For hyperscalers like Microsoft or Google that have already been experimenting with load flexibility commitments, this could create a replicable model. 

Advanced power electronics suppliers. The five reform categories explicitly include "consideration of alternative transmission technologies" in study processes, the first time RTOs have been formally asked to defend whether they are or aren't using them. While not enforceable, if operators actually take this up, it could be an opportunity for ATTs.

Ratepayers, hopefully. Cost-shift prevention is one of the five named reform areas. Whether it produces real protection depends on what the RTOs file.

Data centers without power secured. Projects that are announced but not yet under construction and don't have a clear power strategy are the most exposed. Per Currence’s own data, a meaningful share of the 12 GW of US data center capacity announced for 2026 has no disclosed power plan. The adequacy reports due next month could make that gap very public, very fast.

RTOs. What they say in these reports matters. It’s not clear how much coordination there will be between them, or how forthcoming they’ll be if they do share notes. But their responses will position them as more or less ready to connect large loads and will establish a comparison or ranking between them and against other regions, intentionally or not.

A quick read on the numbers shaping the market. The capex, the contracts, the regs, all anchored in the so-what.

75MW+ // Projects in ERCOT's new approved Batch Zero large-load interconnection framework. It’s the first batched interconnection process for large loads at any US ISO. ERCOT will group all of these 75MW+ projects that have secured financing and land into a single study, rather than processing them in sequence (the model that created the national interconnection queue backlog in the first place). Of course, that backlog is 438GW, 90% data centers. ERCOT will review simultaneously to assess aggregate transmission needs, allocate grid capacity fairly, and identify required upgrades before any project is approved. ERCOT expects to notify Batch Zero participants of their classification in Aug 2026 and finalize the plan by fall 2027.

$17.5bn // DOE loan commitments to finance 10 Westinghouse AP1000 nuclear reactors. The agency says it’s allocating pre-purchase financing for pressure vessels, steam generators, and main coolant pumps to five eligible projects, from unnamed utilities and energy companies. If projects can qualify, this could backstop costs to get the long-awaited nuclear fleet built.

2.67GW // Chevron and Microsoft’s deal for co-located gas plant and data center. Despite Microsoft’s sustainability goals, this is one of the biggest planned co-located gas plants and data centers (aiming for power in Texas by 2028), with the majority of generation via two large GE Vernova turbines. Oil majors are officially entering the hyperscale game, with Chevron brings West Texas acreage rights, gas supply certainty through its upstream operations, midstream access, and a big balance sheet. 

$6bn // Cost of transmission line to deliver Canadian hydro to NYC. The 339-mile Champlain Hudson Power Express is North America's longest fully-buried transmission line, privately funded and developed by Blackstone and TDI, took 16 years to finish. Power deliveries began this month, set to delivers up to 10.4 TWh annually, covering approximately 20% of NYC’s electricity demand. It’s the most significant new supply source for the region in a generation.

Explore more Signals on Currence here.

The policies, rulings, and company moves worth watching.

PJM's 2028/29 Base Residual Auction opens next week. Per JPM’s request, FERC having reinstated the price collar in April, capping clearing at $325/MW-day with a $175/MW-day floor. The collar's return matters: without it, the auction was structured to run with a $550/MW-day cap, and given three consecutive auctions clearing at or near their ceiling, an uncapped result would almost certainly have set a new record.

DOJ moved to dismiss the lawsuit challenging xAI's 46 unpermitted gas turbines. The government argued that Grok is one of four AI models running on classified military networks, so the turbines qualify for a "national, economic, and energy security" exemption from Clean Air Act enforcement. If that survives judicial review, any hyperscaler with a government cloud contract could have a precedent to bypass standard air permits on behind-the-meter generation.

TVA's preliminary 2026 IRP projects up to 26GW of new gas by 2040. The range spans 7GW (Reference) to 26GW (Higher Growth Economy driven by AI and data center load), and actual demand is already tracking above the Reference case thanks to data center growth across Tennessee, Alabama, and northern Mississippi. TVA also sits on 6GW of SMR agreements with GE Hitachi, X-Energy, and TerraPower, making it a potential candidate among the unnamed utilities in the DOE AP1000 loan program.

The latest research, features, and data drops on the Currence platform.

Acario Innovation, the corporate venture arm of Tokyo Gas, has to spot the companies and projects shaping the future of energy before the rest of the market does. To keep pace, the team turns to Currence to see where the data center buildout is heading and who is winning. "Currence shows us the power strategy and specific vendors for a wide array of projects. This is incredibly valuable," says Nigel Carr, Director at Acario. Read the full case study.

Our friends at Energy Central are running their inaugural workforce survey and want to hear from the power sector. Take their survey in just a few minutes -- have your say on hiring trends, AI readiness, and the talent challenges ahead.

Interested in diving deeper? Talk to our team and leverage the tactical intelligence that hundreds of energy and investment decision-makers like Southern Company, Tokyo Gas, Jefferies, Galvanize, B Capital and others use to stay ahead in the energy transition. 

Got forwarded this email? Subscribe here.