Newsletter 065

Happy Tuesday. Before the calendar runs away from us, we wanted to take stock of what we got done in the third quarter of 2026. And Dan's back, twice: on “how we learned to stop worrying and love low Earth orbit,” & providing his firsthand Blackbird perspective.

IN THIS WEEK’S EDITION

⚡️ The EMP problem hiding in low Earth orbit
🧾 Q3 report card: flippenings, first concrete, and more
🐦‍⬛ The man who grounded the SR-71 on bringing it back
🌋 Making steel, hot rocks, and dual-use deal-making
🎨 One more thing: bringing beauty back…

Forwarded this? Join Per Aspera today.

Dan here. I spent 25 years at TRW working on classified national-security space programs. One of them was Milstar, a communications system built to keep the President connected to US forces even after nuclear detonations in the atmosphere or low orbit. Survivability was not an aspiration tucked into a proposal. It was the job.

Milstar operated in geostationary orbit, roughly 22,300 miles above Earth. It was hardened against electromagnetic effects, shielded against radiation, able to alter its orbit, and designed to keep operating for up to six months if its ground-control centers were lost. We knew how to build for the unthinkable. We built it.

Now we are putting more and more of America’s communications, sensing, commerce, and national-security capability into Low Earth Orbit. LEO is extraordinarily useful, and we should keep building there. But if too much of what matters depends on one orbital environment, an attack there becomes a problem for the whole country, not just for the satellites that are hit.

The answer is not to retreat from space. It is to build an architecture that can take a punch: different orbits, redundant paths, hardened spacecraft, and ways to keep operating when a constellation or its ground systems go down. Above all, we need to buy survivability, not merely more satellites.

Read the full article here ↓

What’d You Get Done This Quarter? With Q4 in full swing and some laptop warriors already sending “can we circle back to this next year?” emails, we want to reflect on what our people (the hard pursuits crowd) got done in Q3. We’ll start with a view from the top and work our way down from there.

MACRO: Factories ran a little hotter while Wall Street marked hard tech down.

GROUND BREAKING: The quarter’s biggest concrete pour was Micron’s, at its $100B Clay, NY campus on July 9, a quarter ahead of schedule, and the memory maker boosted planned U.S. spend to $250B+ through 2035. Not to be outdone, SK hynix started building the nation’s first HBM packaging line in Indiana ($4B). Urenco broke ground on a 50% expansion of the only commercial uranium enrichment plant we have, in Eunice, NM, and USA Rare Earth ($1.2B, South Carolina) and Energy Fuels (Utah) went to work on magnet and heavy rare earth plants. The Army and Czech-owned MSM kicked off a 155mm line in Iowa rated for 36,000 shells/mo, less than a year after a contract was inked, and Avio put the first shovels in for a $500M solid rocket motor factory in Virginia. Last but not least, Saronic delivered a Texas-sized buzzer beater on Sept. 30 when it broke ground on Port Alpha down in Brownsville, TX.

GREEN LIGHTS: In July the FCC threw out the rulebook American satellites have been licensed under for decades and replaced it with the Part 100 "assembly line," in August the White House rewrote national launch policy to ready the ranges for 1,000 launches and reentries a year by 2030 (#062, #063), and on Sept. 30 the FCC voted launching and operating spacecraft out of NEPA's "major federal action." On the ground, the NRC handed TVA its Clinch River permit four months early, the first for a commercial SMR in the U.S., and an Arizona zinc-manganese mine became the first to clear FAST-41. And DOE picked 31 SPARK grid projects ($1.9B federal, $3.35B matched) to squeeze 23 GW of new transmission out of wires already in the ground (no new right of way!):

POINTS ON THE BOARD: Five test reactors have gone critical under DOE’s pilot program since June, two of them in Q3: Aalo’s on July 4th, and Oklo’s in August in Lockhart, TX. Anduril’s Arsenal-1 rolled out its first Ohio-built Fury 557 days after the factory was announced (while Austin-built Saronic ships logged the first combat strike by American sea drones). And in a theatrical finish, on Sept. 28, Starship reached orbit and deployed 26 Starlink V3s.

Let Starship’s journey serve as a reminder to always keep pushing. Now, here’s to finishing out the year strong! 💪

Like any aerospace engineer or any fan of going faster, we got a little giddy last week with all of the Blackbird SR-71 hubbub (#064). So we asked y’all: should she (or something like her) fly again? Your verdict was a resounding yes, with a strong preference for the sequel over the remake:

🚀 Yes, build what comes next: 83%
🐦‍⬛ Yes, fly the original again: 14%
🏛️ No, let her rest: 3%

Another reader was up in an airshow control tower filing a flight plan when an SR-71 pilot radioed for clearance to climb to 85,000 feet, roughly twice an airliner's cruising altitude. The tower's reply: "If you can get there, you can have it!" The pilot lit the afterburners, pulled the nose up, and the jet was gone before our reader had "caught my dropped jaw."

That same reader invoked Faster, Better, Cheaper, which brings us to the man who made it famous. Our own Honcho, Dan Goldin, was running NASA the last time the SR-71 flew, and decided to take us up on our solicitation for Blackbird stories.

In 1999, I pulled the SR-71 from flight status. It wasn't an easy call. I love that airplane. Every engineer does. NASA had been flying Blackbirds as research platforms since the early '90s. By the end of the decade, the research we could still get out of her didn't justify what it cost to keep her flying, and we had other hypersonic work we wanted to fund. Her last flight was nearly 27 years ago to the day (October 9, 1999).

People forget how precisely she was built. The SR-71 is a tuned Swiss watch designed to do one thing: go just a skosh beyond Mach 3.1. Change anything, however small, and it ripples through the whole machine. When we flew her, we still had spare parts, engines included. Any modern take on the Blackbird idea will likely need new structural parts, and possibly a new airframe altogether. It's hard, expensive engineering: exactly the kind of hard engineering this country is best at. That's why I get excited watching today's X-plane conversation.

Mach 3 was never where I wanted us to stop. I wanted Mach 10, AKA, kick-a** hypersonic speed. We got close. In November 2004, NASA's X-43A scramjet hit Mach 9.68, and it's still the fastest air-breathing aircraft ever flown.

Kelly Johnson and his Skunk Works team built the Blackbird with slide rules, and the best tribute to that genius is building the next impossible thing. The good news is the field is alive again: Stratolaunch's Talon-A has now flown past Mach 5 more than ten times.

So when NASA says it wants to fly faster, count me in. I've been waiting a long time to hear it. 🇺🇸

001 / BIG BOYS FIRST... Some kinds of steel are getting hard to come by, with smaller shops feeling the squeeze first. One metals buyer posted that mills are now quoting three to six months out and “starting to ration (big boys first of course),” with a machine-shop owner replying that it’s bad “even for small round bar.” The macro numbers back this anecdata up: steel prices are up ~23% on the year, with datacenter construction rising 57% in July and factory construction falling ~22%. A rising tide, mostly, but this is fresh evidence for something we’ve worried aloud about for a year (#029, #062): in the tightest corners of the market, is the price-insensitive AI supercycle crowding out other critical industrial efforts? Some additional perspective: U.S. mills are running at ~78% of capacity, and 50% tariffs have kept most imported steel out, so policy and product mix have partly conspired to form the squeeze. And new supply is on the way: Nucor’s new West Virginia sheet mill will finish construction this year and ramp through 2028, Hyundai Steel’s $5.8B Louisiana mill is targeting 2029, and last week, Mesabi Metallics pledged $18B for a 10M-ton-a-year complex in southeast Iowa, fed by what it bills as the first new U.S. iron ore mine in 50 years. First steel there is due in 2030.

📣 REQUEST FOR COMMENT: Buying or selling steel right now? We want to dig into this one further and hear from those of you running a mill, service center, factory, or job shop. What are lead times, allocation, and pricing looking like from where you sit? Is this indeed a big issue from your POV, or much ado about nothing? Hit reply or click below and let us know.

002 / HOT ROCKS, A DAY EARLY… On Sept. 30, Fervo Energy’s Cape Station in Utah entered commercial operation one day ahead of its contractual deadline, making it the first utility-scale enhanced geothermal plant on the grid. Conventional geothermal needs hot rock, natural cracks, and a natural underground reservoir of water to carry the heat out, a rare combination that helps explain why the method is confined to a handful of Western hot spots and makes just 0.4% of U.S. power despite being around since 1960. Fervo’s trick was to leverage technology from the shale revolution: drill horizontally, fracture the rock in stages, pump water through, and watch every well with fiber-optic sensors. In other words, it only needs the hot rock, which opens the map for far more viable sites and, like shale, should improve unit economics with every well. With the publicly traded Houston company’s first 33 MW block online, the next test is two more blocks for a total of 100 MW by Jan. 1, then another 400 MW by 2028. Other startups are chasing the same heat with different tricks: Mazama is drilling for superhot rock (#027), Eavor circulates fluid through sealed closed loops, and Quaise wants to melt its way deeper with millimeter waves.

003 / DOUBLE OR NOTHING... Two defense autonomy makers raised big rounds in September. Mach Industries, founded in 2023 by Thiel Fellow Ethan Thornton, closed a $600M Series C extension led by Ribbit Capital, Infinite Capital, Bedrock, and Sequoia at a $3.7B valuation, double its valuation three months prior. Mach builds airframes, jet engines, solid rocket motors, and energetics in-house in its 115,000 sq ft Huntington Beach plant. In Lisbon, TEKEVER raised $580M in the first close of a Series D led by UC Investments and Baillie Gifford at $6.4B (up from ~$1.3B a year ago), fresh off a UK MoD deal worth up to £400M over ten years to replace the British Army's Watchkeeper drones.

001 / DID WE SAY: WHO CONTROLS THE AIR??… Congrats to SoloPulse, seed partner of our friends at OOTWD, on landing a new backer: Dajji, the deep-tech platform founded by former Nokia strategy and technology chief Nishant Batra. The Atlanta startup, a Georgia Tech spin-out (🙌🙌🙌), forms a radar image from a single RF pulse. As we’ve said before (#047): everything is spectrum.

002 / ANTARES, FLIGHT-BOUND… Antares (s/o cofounder Julia DeWahl) has landed a $161M Air Force award to ground-test its R1-S space reactor and certify it for flight. Antares calls it the largest current DoW space-nuclear award, and it lands just months after their Mark-0 went critical at INL (#057) — the first private advanced reactor to do so under the DOE pilot program (can you believe that was just a quarter ago?! progress moves quickly!).

003 / JPMORGAN GOES HYPERSONIC… JPMorgan's Strategic Investment Group co-led Castelion's $1B Series C at a $13B valuation alongside a16z and Carlyle, putting one of America's biggest banks directly behind a hypersonic-missile maker. It follows JPM checks into Shield AI and Helsing, and gives teeth to Jamie Dimon's Security and Resiliency Initiative, which commits up to $10B of the bank's own capital to strategic industries.

AMERICA THE BEAUTIFUL & GLAMOROUS...

👋 Joy Shin here (PA Cofounder). A few days ago, an X user who goes by Dume posted Detroit's Fisher Building.

The seven Fisher brothers began as carriage makers, then built Fisher Body into the company that popularized the enclosed automobile body — which made year-round driving possible.

In 1927, they hired Albert Kahn to give Detroit a headquarters equal to the machine age: a thank-you to the city that delivered their success, an ode to American work and American artisans, and — by their own ambition — the world's most beautiful office building. It opened 15 months later as a 441-foot Art Deco tower, with a hand-painted, barrel-vaulted public interior of marble, brass, bronze, mosaics, frescoes, and sculpture.

There is a lot of good talk about building beautiful things. But perhaps the distinctly American extension of beauty is glamour.

Glamour originally meant enchantment: a spell cast over what the eye sees. Technology does the same in material form. It takes difficult, specialized knowledge and turns it into cars, towers, radios, rockets, and other things ordinary people can see, use, inhabit, and be excited by. The Fisher Building made the automobile age visible, turning what happened inside factories and on balance sheets into something Detroiters could walk into.

For people around the world, America has long held that kind of glamour: the sense that the future is being made, and that it might include you.

But a glamorous society requires more than taste. It requires a country able to build quickly and durably; efficiently and with craft; for private return and for public life. Authoritarian systems can build monuments. The harder achievement is a country where ordinary builders compete to make ordinary life more extraordinary.

Here's to all you glamorous builders making America feel magical.