👋 It’s me, Dan Goldin! It’s been a while since my editor Ryan handed me the reins, but I’ve been cooking on an idea. And so today I have a big thought across two stories for you all today:
The present-to-future story: My reaction to Elon Musk’s Terafab speech this weekend. I think it was a BFD, in all of the right ways, and not because of the talk of mass drivers or space-based datacenters….
The past-to-present story: A patent nobody in Washington had heard of, an 11,000-pound problem, and the weld that flew 100 times without failing…
My big thought: great engineering happens when teams are cornered by conventional solutions. To me, when America is cornered by a structural problem, the escape is always through people doing good physics.
So here’s to you all, in the business of good physics. 🥂
IN THIS WEEK’S EDITION:
⚡ Terafab is forcing the right conversations
🔧 The unsung hero that saved the space station
📀 The stat: a $1T allied supply chain fundraise target
👩🌾 Threats, Moonbase, and American Farmers
🏰 Jeff Bezos is on a roll(-up)
Forward this newsletter to someone in your life who’s in the business of good physics. If you were forwarded this, Subscribe to Per Aspera here.
It's been 64 years since a younger, more idealistic Dan Goldin heard JFK say we go to the Moon not because it’s easy, but because it’s hard.
No speech has moved me like that one — until I saw Elon take the stage in Texas this weekend and announce Terafab.
The case for Terafab
Tesla, SpaceX, and xAI intend to build the largest semiconductor fab ever attempted, in Austin, at the most advanced node on Earth. We break out the full specs and the details of consequence in the published piece — but there is a clear case for building this, and it goes well beyond the space datacenter thesis (though we've thought about that plenty).
What stopped me cold was not the logic for building one’s own logic chips, nor the seemingly long odds.
Reading between the lines
What moved me was watching an American team propose, for the first time in 30 years, to build a first-of-a-kind, leading-edge logic foundry in the United States of America. To build something new that doesn’t merely copy-paste what already exists nor work off of a narrow scope of existing processes, templates, and architectures.
Instead, what I saw was Elon going back to the physics and first principles and asking: Does it have to be this way?
Could it fail? TSMC spent $100Bs and generations accumulating the tribal knowledge to print transistors at 2nm. The companies in the Musk Cinematic Universe have yet to ever manufacture their own chips. Jensen says it’s extremely hard. Of course it could fail!
Someone has to try
The idealist in me — the kid from the Bronx who heard JFK, threw away his grad school application, and walked straight into NASA — wants all of it to work. The jaded realist who spent a decade running the agency knows it might not. But why not try!? And if not SpaceX and Tesla trying to tackle on closing this two-sided gap (both sovereign capacity + demand-side), then who?
Even if the full vision falls short, or the space-datacenter thesis fails to play out, the byproducts of a true Terafab program could be well worth it: a sovereign American fab, new compute architectures, rad-hardened silicon for national security, and so on. I have six questions I'm watching and one warning I felt obligated to raise (hint: look to the East India Company)…
…but overall, I have the conviction that this speech was a BFD. Keep reading for exactly why.
And now for a story on friction stir welding (FSW). Or as we like to call it, the Unsung Hero That Saved The Space Station.
The story goes a little something like this: back in 1991, a metallurgist named Wayne Thomas in Cambridge filed a patent for a process that joined metal without ever melting it.
Little did Wayne or I know, but the process he had pioneered was about to save our bacon station.
Across the pond, shortly after this patent was filed, we struck a diplomatic agreement in Washington with the Russians. We’d build the ISS together, and put it in an orbit that let the Russians launch from Kazakhstan. Only one problem…it left the Space Shuttle 11,000 pounds short. The Shuttle had to lift the components to assemble the station. Without those 11,000 pounds, it could not do the job.
So how was I feeling?
“I was shi*ting in my pants.”
Keep reading for the story on FSW — from the scramble to shed weight, to the alloy that could save the day, to the engineers who made it work, to the proliferation of FSW from rockets to shipyards to EV factories worldwide. We should be using it in more damn places!
One trillion dollars is the aspirational investment target for Pax Silica. The White House’s consortium designed to channel allied capital into supply chains and bundle chips, minerals, and energy under one allied umbrella. The State Department is putting skin in the game with a GP seed check of $250M and assembling a team, with SoftBank, Temasek, and Mubadala, who oversee $1T+ in AUM, among the founding members of a voluntary investor group spanning 13 allied nations. Jacob Helberg, State’s point man for Pax Silica, framed the urgency of the initiative through the lens of the Strait of Hormuz:
“We are witnessing how a single bottleneck in the global economy can be weaponized. We will never tolerate such calculations spreading to technology and supply chains.”
Though there’s an irony here — the same crisis that makes Pax Silica’s case may also complicate its funding. Gulf sovereign wealth funds are among the world’s largest pools of dry powder, and a region at war tends to turn inward, not write fresh LP checks or channel funds into new builds halfway around the world. And, for context, total global FDI last year was $1.6T — so the $1T target is ambitious by any measure. Per Aspera will always cast our support to ambitious projects with Latin names. (Pax Silica: “peace through silicon.”)
001 / THE IC’S PHYSICAL … Each year the nation’s Intelligence Community (IC) synthesizes and consolidates the work of its 18 agencies into an annual threat assessment report. A five-point breakdown of the unclass version, presented to Congress last week:
The homeland faces a five-fold increase in ballistic missile threats from ~3,000 today to 16,000 across five countries by 2035. Everyone’s talking about counterdrone technology, and they should be, but let’s not forget the sustained, multi-nation missile production surge that will test every layer of defense we have or plan to build.
On that note, the IC says Beijing already fears Golden Dome will lower America’s threshold for using force in a crisis, which is pushing China toward arms control talks around the proposed system’s space-based elements (creating a sort of recursive deterrence before anything is even fielded).
The Arctic is a place to watch. Russia is concentrating second-strike capability and its massive icebreaker fleet there. China, a non-Arctic country, is working to establish a presence there and pushing to fold a "Polar Silk Road" into Belt and Road as shipping lanes thaw.
On Taiwan, the IC assesses that China won’t invade Taiwan in 2027 and that no fixed timeline exists, though the PLA keeps building relevant capability.
In a narrative violation, the IC says “adversary alignment” is an overstated phenomenon, and that coordination between China, Russia, Iran, and North Korea is “limited and primarily bilateral” and constrained by divergent interests.
It’s not the rosiest of reads but a clear-eyed one, and probably pound for pound about the best you could do in insight density per page in 34 pages of geopolitical analysis. Worth thumbing through if you’re after an intellectually heftier alternative to doomscrolling.
The astronauts of Artemis II, who could launch on their crewed lunar flyby mission as soon as next week.
002 / MOONBASE … Jared Isaacman is making clear and decisive the priorities for our national space program. Yesterday, the NASA boss unveiled “Ignition” at HQ, a package of directives to execute the National Space Policy: a phased plan for a $20B permanent lunar base, crewed landings every six months and up to 30 robotic deliveries by 2029, an effective cancellation of the Gateway station to redirect resources to the surface, and — after decades of study — the launch of Space Reactor-1 Freedom, the first nuclear-powered interplanetary spacecraft, to Mars by EOY 2028. As Isaacman said: "The clock is running in this great-power competition, and success or failure will be measured in months, not years." (PS for students/researchers/startups: via a new RFI released yesterday, NASA is offering your ideas and payloads the chance to reach the lunar surface through CLPS commercial landings that will ramp up starting next year.)
003 / “AN OFFER THEY COULD REFUSE” … So read the chyron for a local news story on a Northern Kentucky family who rejected $26M from an unnamed Fortune 100 AI company for half of their 1,200-acre, generational farmland (which is 10× nearby acreage’s going rate). This is at least the third such story to go viral in the last couple of months — see our recent story about Mervin Raudabaugh in Supermarket of the World — and it already has tens of millions of impressions, and overwhelmingly sympathetic sentiment, online. The instinct in some tech circles is to sneer/dunk on these landowners as sentimental NIMBYists, neo-Luddites, anticapitalists, or some combination thereof. What the tech commentators fail to see, and what Big Tech leaders and the White House alike understand, is that datacenters/AI have a big PR problem. AI carries a net favorability of -20 among registered voters, which makes it slightly less popular than ICE but slightly more popular than … Iran.
Eagle-eyed readers out there will recognize that this graphic is a variation of one we ran last week — just with Bezos up top this time. While we didn’t intend to run this in back-to-back editions, the Amazon, Blue Origin, and Project Prometheus founder gave us no choice. Reports surfaced last week that Bezos is channeling his inner Masayoshi Son and pitching investors on a $100B fund to acquire and AI-ify manufacturing companies across chips, defense, and aerospace. As we’ve just been saying: You just can’t sit this one out!
|