Powerstack: Jesse Jenkins on the fastest clean firm(a) power

It was great to see so many of you on Monday night! Thanks for coming to our Energy & AI Rooftop party for the great weather, view, and vibes. Plus, the big reveal: Sightline Climate is now Currence, market intelligence to power the AI era.

And we’re hosting another great event in London next week on 23 June, what we’re calling Building & Financing AI in the UK. We’ll share the latest Currence analysis, and hear from Ofgem, Octopus, GE Vernova, and Deep Green. Space is limited , but register here.

Below, we've got an interview with Jesse Jenkins, of energy Twitter fame and now, Firma Power founding. Watch the whole version on our YouTube here, and read on for highlights.

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

Jesse Jenkins leads the ZERO Lab at Princeton, where he's an associate professor of energy systems engineering. He's also CTO and co-founder of Firma Power, a new startup assembling clean firm power portfolios for data centers. We got intel on the new company, his bull case for getting fast clean firm power for large loads, his contrarian take on where data center demand is actually headed, and more. The Q&A below is edited for length and clarity, but catch the full interview on our YouTube here.

On if all the data center demand is real

On the demand side, yes. Jenkins puts 60 to 80GW of additional load growth by 2030 in credible territory. But the question is if supply can meet it. “My suspicion is that a lot of that demand is at risk of being pushed out because we're not able to build fast enough on the power side. The broader power sector is not set up for that right now.” 

Power infrastructure has long lead times, generation and transmission. Projects mature enough to build now started development 3+ years ago when data centers were not the objective. They haven’t been developed to serve this load. Anything you want to start now, greenfield, to respond to this, won’t be ready for another 4-5 years, so already to 2030 or beyond. “That creates this really unique challenge between now and 2030, with tens of gigawatts of load growth per year, but you don't have the whole system, organized and working to meet that demand growth. In fact, very much the opposite.”

"The system was organized for flat demand and to scale wind and solar as energy and RECs supply for corporate C&I customers, not as capacity-enabling, firm resources. Data center developers used to just go to the utility and say, ‘Hey, I got 50MW, can you connect me?’ And they'd say, ‘Sure, you know, we'll connect you in 18 months so go ahead and we'll just add you into broader demand and we'll find supply for you.’ Well, now you're showing up with, you know, 500, 800, 1,000MW or GW of request, and the utility, especially the smaller mid-size ones, you know, like rural co-ops that might have a peak demand of 2GW. They're saying, ‘Well, I can give you 50 or 100MW, but the rest is up to you to figure out.’ So now people are scrambling to find alternatives.”

On the real options for clean firm power - fast

Restarting Three Mile Island, next-gen nuclear, advanced geothermal: all real, all post-2030. The fastest clean firm megawatts are the wind, solar and storage already deep in the interconnection queue. Roughly 98% of what's in the queue is clean. It just wasn't built with data centers in mind. That's Firma's thesis, Jenkins says.

The right resources are scattered. Firming a 1GW data center takes something like 3GW of wind, solar and storage in the same deliverable region, and no single developer has that mix in one place, not even the big ones like NextEra or Invenergy. The point used to be to have a geographically diverse portfolio of projects, but that doesn’t work for a capacity-enabling transaction like for a data center.

Firma solves two problems: the analytics of which portfolio of in-development and/or operational assets actually delivers the accredited capacity (which varies by jurisdiction), and the contracting to pull five to eight projects across four or five developers into a single offtake. It's a hybrid model. Buy some assets pre-NTP, contract others via PPA, bundle, and sell one contract through to the data center. [Learn more in our full interview here.]

On how it all compares to just building gas

There isn't enough gas to meet the demand alone, so Jenkins doesn't treat it as the direct competitor. And zooming out, islanded behind-the-meter gas won't even necessarily retire into backup. Once a 400MW operator finally gets its grid connection, the economics say double the data center and keep running the turbines. "I'm very skeptical that those resources will ever shift into a backup role,” Jenkins says.

On other creative solutions

Flexible interconnection has big potential and real bottlenecks. Utilities don't run hour-by-hour 8760 modeling, so they can tell you they need 300MW in one stressed moment, but not whether it's for four hours or forty, or once a year or 15 times. Without that data, "that makes it pretty hard for you to size your on-site resource."

What's needed is for utilities and RTOs to invest in a more dynamic understanding of grid constraints. That would expand the on-site toolkit: compute flexibility (as Emerald AI, where Jenkins advises, is pursuing), cooling system flexibility (ZERO Lab’s researching), or on-site power and storage. "You have to really know what you're trying to solve for, and we don't really have a good answer from the utilities on that right now in most cases."

And edge compute will be a slice, the way distributed energy resources are, but the centralized backbone isn't going anywhere.

Most contrarian take

That "gas now, nuclear later" is the default - it misses a huge scalable option. Clean firm portfolios can deliver tens of gigawatts in three to four years, now, when they’re constructed right. And that’s while pulling forward demand for nuclear and geothermal in the 2030s.

The call to action

Rebuild social license. With 70% of Americans anti-AI and data centers, Jenkins says the industry is doing it wrong: secretive siting, polluting on-site generators, tax breaks, no local upside.

“You have to have a better pitch than we're gonna build 60 jet engines in your backyard, run them all the time, take all your water, and then power data centers, whose job is to take all your jobs.”

Clean quiet power solves the air and noise piece. The harder work is giving communities and the public a real stake in the AI economy.

Mark’s take

I definitely had a favorite moment in this interview.

It was when we mentioned our work on The Fastest MW, and Jesse said something to the effect of, that report certainly did not go unnoticed by my co-founders and me. 

I’ve been a fan and follower of Jesse’s work for quite some time, and great to hear that we independently arrived at a very similar conclusion. That the fastest route to capacity is not gas now and nuclear later, but it’s actually later stage renewables and batteries in markets like ERCOT. And to hear he's now building a company on that same conviction – that was a cool moment.

With Firma Power, Jesse and his co-founders are after something bigger, a question we never set out to answer. Our Fastest MW work mapped which clean assets come online quickest. Firma works out which of those can be bundled into firm power for one data center, then sold through as a single PPA.

I love any strategy that is truly creative and is an attempt to hack at the roots of a problem, and this is one of those. We’ve written before that one way to get capacity faster and compete in this new gold rush is to buy the platform, or, buy a portfolio of assets in development and essentially control your own destiny – see the Google/Intersect deal, or even more on the nose, GIP/EQT buying AES. Firma's move is a clever sidestep. It doesn't buy the platform at all. It assembles a synthetic one through analytics and contracting, no billions required (although it sounds like they might have some plans for that, too). To me, the magic is in the analytics, knowing which specific assets, in which region, deliver the accredited capacity a data center actually needs.

The danger is in the roll-up. One PPA here means five to eight projects across four or five developers – that's a lot of hoofin’ it for signatures. But that's also where this gets interesting, because if I’m looking at this, Firma doesn't have to assemble the portfolio cold.

If I'm a banker, an infra investor, or a project-finance lawyer reading this, my ears are piqued. Data centers are getting financed off balance sheets and gigantic equity raises, think Google and Meta, but the solar, wind, and storage underneath is still project finance, and a steady stream of those deals already crosses my desk. But what I don't know is which of them belong together in a portfolio like this. Partner with Firma and I get the answer, and access to the roll-up. This bulk approach could reduce the signatures problem and supercharge the go-to-market.

What we’ll be watching is the approach Firma ultimately takes – to be the ones doing the roll-up, or to be a software provider everyone else uses to do them. 

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

250MW // FERC approved PJM's Expedited Interconnection Track, creating a channel to fast-track 10 interconnection requests for new or uprated capacity. Projects must have a 250MW minimum, $500k study deposit, $15k/MW readiness deposit, 10-month timeline to GIA, effective July 31. But it sunsets December 2027, by which point the next process, PJM’s reformed Cycle Process, could scale, if all goes well. 

3 // Rolls-Royce SMRs selected to deliver Sweden's first new nuclear build in over 40 years. It’s three units, ~1,500MW total, targeting a west coast site with the first unit online in the mid-2030s. Rolls-Royce beat 75 competitors, including GE-Hitachi, which entered the process as the leading US/Japanese option. With GE-Hitachi having a much more mature product there are a range of possible explanations, from price to terms, it’s interesting that Sweden went with the European solution amid rising energy security concerns.

$16.6m // BLM’s geothermal lease sale in New Mexico for 158,000 acres. Pricing hit $108.82/ac on average (but winning bids varied by parcel), matching Utah's 2025 record. The per-bidder breakdown: 6 brand-new LLCs took 60% of the acres (Novo Mesa, Rock Canyon, EDM Geothermal, Magdalena Energy, Entrada Energy, and EGX Energy, none of which have web presences nor prior BLM history). Could be stealth or new entrants. Either stealth positions by named operators (Aspect, Fervo, XGS pattern) or new entrants. Established players like Ormat, Invenergy, and Zanskar also all showed up.

$765m // what Interior will pay Invenergy to surrender four offshore wind leases. The deal pushes the running tally the administration has spent buying out offshore leases past $2.5bn, following earlier payouts to TotalEnergies and others. Invenergy is steering the cash into at least five gas plants across Indiana, Wisconsin, Iowa, Kansas and Missouri, plus geothermal out West, for the first time in the wind buyout saga.

45.5TWh // Solar generated more electricity than coal in the US for the first time. A new Ember report showed solar was 12.8% of the mix vs. coal's 12.2%, as coal's share has nearly halved in five years from 19.7% to 12.2%. Mild May weather isn’t the main reason, it’s that utility-scale solar has grown 20% year-over-year, despite the DOE spending $850m on new coal plants the same month.

Explore more Signals on Currence here.

The policies, rulings, and company moves worth watching.

FERC's June 18 open meeting vote on RM26-4. The large-load interconnection rulemaking, which the Trump administration directed the agency to make, is on the agenda. SPP CHILLS (non-firm large-load access with curtailment exposure, approved June 5) and PJM's Expedited Interconnection Track (approved June 9) are the recent precedents FERC is building on. Depending on scope, RM26-4 could range from a targeted NOPR to a full proposed rule with immediate market implications for PJM, MISO, and CAISO tariff negotiations. 

Form Energy has reportedly tapped JPMorgan and Jefferies to lead its IPO, with Goldman and Morgan Stanley competing for the third slot, the clearest sign yet it's headed for public markets. With valuations running hot and 75 GWh-plus under contract from Google and Crusoe, the iron-air batterymaker has the backlog to make the case, even as its first real revenue only lands this year.

The bipartisan FREEDOM Act, the energy permitting reform bill. Sens. Tom Cotton (R-Ark.) and Catherine Cortez Masto (D-Nev.) introduced the bill promising "permitting certainty for energy projects across America, without favoring any specific technology,” so technology-neutral framing that covers fossil fuel infrastructure alongside renewables and transmission. While it’s not out of committee yet, it’s another shot at permitting reform after the last bill failed in the Senate because of its more explicit pro-renewables framing.

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

Introducing Data Centers & Power: Tracking project credibility for the AI buildout. A new product from Currence fills the missing layer in data center intelligence: project-by-project credibility on the AI buildout, so utilities, financiers, and suppliers can make better bets on what gets built. Learn more in our post here, or catch us in real time at our webinar on Jun 25. 

Where the market is meeting, and where to find us

📅 Building & Financing AI in the UK // London, 23 June, 2026 // Held during LCAW, join us for our flagship London event. This year we’ll focus the discussion on the data center buildout (what else?!) and the pathways to the fastest MW. Mark your calendars! Register here.

📅 Quantifying the cost of data center delays // Virtual, 25 June, 2026, 10am ET // We went viral for our data center delay analysis. Now we're going to find out what it will cost. Register here.

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.