Good morning folks! We talk endlessly about the source of energy to safely, efficiently and affordably power our data centers and industry growing needs, but the United States still has no serious public–private framework to move fusion energy, a most potent solution to our need from experiment to the grid.
Since my boyhood, the promise of fusion energy as the holy grail has lingered — the promise of an inexhaustible, low-carbon powerhouse with far less long-lived waste and proliferation risk than the fission reactors that triggered public outcry at Three mile Island, Chernobyl and Fukushima.
And today, I see fusion as the decisive factor for our American AI dominance — a requirement that should matter to you whether you work in industry, education, defense, or anything touched by computation.
I luckily get to call Ed Moses — one of our nation's fusion leaders who ran the National Ignition Facility and now leads Longview Fusion — a friend, and naturally, I pulled him in to help us articulate what it will actually take to move fusion from experiment to the American grid.
Thank you Ed, for contributing to our Per Aspera community here. Let's go!
IN THIS WEEK’S EDITION:
⚡ Fusion: core power
🫵 BCI Survey results
🥈 Supercycle winners
🙌 Friends in high hard places
Forwarded this email? Subscribe to Per Aspera here.
There’s an old joke in energy: “Fusion has been 30 years away for 60 years, and always will be.” It’s so common it’s on Wikipedia.
You hear it less these days. Three years ago, the National Ignition Facility at Lawrence Livermore focused the world's most powerful laser on a peppercorn-sized fuel pellet and created a miniature star — one that produced more energy than it took to ignite. They’ve done it again since, with better results each time, proving the physics works.
One of us (Ed Moses) led the design and build of that facility, and now runs a fusion startup, Longview, commercializing this technology. The other (Dan Goldin) has spent decades watching the science mature. We wouldn’t be writing this if we didn’t believe commercialization was close.
The demand side
America desperately needs fusion. We’re building the most power-hungry industrial base in our history on an aging, overloaded grid. Datacenters, fabs, defense facilities, and electrified factories are landing across the country and waiting to tie on. The load is growing in every direction, and the grid isn’t ready.
The usual suspects don’t scale, at least not without running headfirst into permitting wars, local opposition, or hard geographical limits. Fusion offers something no other source can match: dense enough to site next to a datacenter or shipyard, clean enough to put anywhere, and powerful enough to stabilize grids rather than strain them.
Just one problem: fusion hasn’t proven itself commercially viable.
The commercial moment
Fusion has lived in national labs for decades. They advance fundamental science, but don’t build commercial power plants. That's the job of private industry…and for a long time, it couldn't answer a simple question: what's the path to grid power? Today, though, the fusion industry is closing in on an answer.
Nearly $10B in private capital has flowed into fusion companies in the last five years, with $2.6B in the last year alone.
We are pursuing a range of technically differentiated paths: tokamaks, stellarators, laser-driven inertial confinement, z-pinches, etc.
Governments in Japan, Germany, China, the U.K., and the U.S. have all passed supportive policies, with public funding up 84% YoY.
We write in today’s essay that fusion feels a lot like the early days of commercial space. While it hasn't yet had its "Falcon moment,” we in the industry are operating as though it’s a matter of when, not if.
What will it take to retire the punchline?
Fusion is a famously hard problem that punishes wishful thinking and breaks the heart of blind optimists. But as we design the grid and gigaprojects of the 2030s now, and with $100Bs flowing into infrastructure, the decisions we make in the next few years will determine if fusion is a real option in our future portfolio — or whether we must lock in another generation of stopgap solutions.
In today's essay, we lay out our full case: why this moment is different, the advantages of fusion over other sources, how industry and Washington can seize the opportunity, and what America stands to lose if we blink. The window to lead won’t stay open forever.
From Y’all…
In Issue #028 last week, we polled you all on whether you’d wear a brain-computer interface (BCI). Perhaps surprisingly, Yes won with 44% of the vote, narrowly beating No at 40% and a healthy Not Sure contingent at 16%.
Here’s what you had to say:
“It would depend greatly on the interface's invasiveness, testing, adoption, and transparency. A BCI whose software is openly-verifiable would go a long way in reassuring the masses.”
“I'm counting this type of technology helping to stave off things like Parkinson's and dementia, both of which are in my family - and possibly my future.”
“It has the potential to increase the throughput between me and the computer by at least 10x, even if this doesn't translate to getting 10x done. It would at least mean 2x getting done. You're no longer bound by your typing speed or mouse clicking.”
“BCI holds the promise of not only an AI-assisted boost of intelligence, a.k.a. the movie Limitless, but it may also be the workaround solution for physical injury and other human ailments where we've underestimated the importance of the mind-body connection.“
Thanks to everyone who voted — keep your eyes peeled for the next reader poll soon!
In American Dreams (Issue #022), we wrote about our once-in-a-generation construction supercycle, largely driven by hyperscalers, who will spend $320B+ on datacenters this year. But we noted that the supercycle reaches far beyond hyperscalers, GPU designers, or frontier model labs — up, down, and across the industrial economy. We already named two second-order winners in Issue #025: the “aircraft boneyard raiders” (e.g., clever aircraft turbine retrofitters) and Caterpillar. Now, it’s time to nominate a third: the skilled trades.
“I pinch myself going to work every day.”
That’s 51-year-old contractor DeMond Chambliss telling the WSJ about his night shift overseeing 200 welders, plumbers, and electricians at a DC build in Columbus, OH. DeMond traded his small drywall business for a supervisor role and now pulls in $100K+ annually. He’s not alone: workers moving into datacenter construction are seeing pay jumps of 25-30%. The money’s great because an all-out arms race to bring new gigawatts online is colliding with a supply-constrained skilled trades market. (Case in point: Becoming a licensed electrician takes 4-5 years; 45-70% of DC construction is electrical work; and 30% of union electricians are near retirement age.)
Skilled tradesmen are the long pole in the tent on every project, so developers are throwing money at the problem and often resorting to extreme measures, such as flying around staff on private jets and setting up “man camps” at remote builds in the Texas Panhandle and the Dakotas. (Man camps = temporary villages that house crews, with dorms, mess hauls, laundry, and full sewer systems.)
Five years ago, you’d start building the second structure on a campus after finishing the first. Now, you build all eight structures simultaneously, run 24-hour shifts, pay overtime, and if necessary, fly trainers around on Gulfstreams.
The Per Aspera POV
The software abstractors have turned into high priests of heavy industry. Under their new industrial logic, competitive advantage comes from building in the real world faster than your rivals. In this game, players are motivated by a mix of paranoia, prisoner’s dilemma, and profit maximilization, and the risk of waiting exceeds the risk of waste. There are certainly critiques one could make about the shape and texture of this renaissance: maybe it’s (A) crowding out other manufacturing, (B) inflationary for inputs for other critical sectors, and (C) the most expensive game of corporate follow-the-leader. But…you wanted an industrial renaissance? You’ve got one, and it’s undeniably reshaping the topology of an entire continent — and crowning new winners in the process.
Breaking/Bonus: A fourth nominee!
While putting the final touches on today’s issue, we saw news from Blake Scholl and Boom Supersonic, which has raised a fresh $300M round. But that’s not the real news. Boom also announced its second product and it’s not another supersonic airliner. It’s Superpower, a 42 MW natural gas turbine optimized for AI datacenters. Crusoe has signed on as a launch customer, placing a 1.21 GW order. The turbines are expected to start entering service in 2027.
JPMorgan Chase has tapped Berkshire Hathaway veteran Todd Combs to lead a new $10B Strategic Investment Group housed within the bank’s recently launched $1.5T Security and Resiliency Initiative. Combs, long regarded as one of the world’s top long-term investors, will step down from his roles as Berkshire investment manager and GEICO CEO to take the job, starting next month. He’ll serve as a special advisor to Jamie Dimon and sit on the initiative’s external advisory council, alongside figures such as Jeff Bezos, Michael Dell, Robert Gates, Condoleezza Rice, and Paul Ryan who will help guide the bank’s strategy across defense, aerospace, critical minerals, healthcare, energy, and more. A big get for JPM indeed!
On December 1, 2025, the Department of Energy rebranded the National Renewable Energy Laboratory as the “National Laboratory of the Rockies,” signaling a shift from a renewables-focused institution born in the wake of the 1973 oil crisis to a broader applied-energy and manufacturing engine for meeting America’s surging power demand. Officials cast the new name as a mandate to build scientific capabilities that restore U.S. manufacturing, drive down energy costs, and secure domestic supply for an all-of-the-above energy mix, rather than “picking and choosing” favored sources. We cited the lab’s research in today’s fusion Antimemo and we’re grateful for the work they do!
Per Aspera Friends in High, Hard Places
Castelion has raised a $350M Series B, led by Altimeter and Lightspeed, signaling a shift from ambitious newcomer to a company expected to deliver real capacity. The startup now turns its focus to moving hypersonics out of the prototype era and into true production. In just over two years, the team went from a blank-sheet design to 25+ flight tests and integration work across Army and Navy platforms, giving them one of the fastest development tracks in modern defense hardware.
This new financing will push Castelion’s Blackbeard closer to full operational integration and support the standing up of Project Ranger, the company’s 1,000-acre solid rocket motor production and final assembly campus in New Mexico. The plant is designed to produce thousands of units per year.
“Blackbeard helps close America’s hypersonic capability gap against China and Russia…This funding lets us build fast, test often, and produce at volumes that matter in the real world.”
Want more? We’d highly recommend checking out Bryon’s interview with Lightspeed, if you’re looking to be inspired…
Last Friday marked the inaugural Global Space Awards, led by Type One VC’s Sanjeev Gordhan — where our own Jeff Crusey served as a judge. A few words from Jeff:
I'm a bit of a curmudgeon when it comes to awards. I think the award is the joy in doing that thing. But I recognize it's important to build culture, and one of the ways we do that is by honoring the achievements of pioneers. It was beautifully marked with the final award, the James Lovell Lifetime Achievement Award — an American patriot and astronaut who flew to the moon twice and survived the Apollo 13 mission. In the end, we all looked back to the American grit and perseverance as the reason any of us got to sit in that room. People like James literally risked his life to move humanity forward. It made me a very proud American that night.
Extra Renaissance Rumblings
SoftBank’s Masa Son is at it again, this time hammering out a high-tech American factory plan with the White House // What the U.S. can learn from China’s success, by Eric Schmidt in The Atlantic // DOE pledges $800M in cost-sharing grants to TVA, Holtec for SMR development // Commerce agrees to invest up to $150M in next-gen lithography startup xLight // ‘26 capacity for major memory chipmakers effectively sold out // An AI program is now MIT’s second most popular major // Apple details how it’s using metal 3D printing at consumer electronics scale.
|