Newsletter 30

Happy Hump Day, folks! It’s that time of the year, where your mobile apps saturate your screen with animations and exclamation points for “X/Y/Z Wrapped.” We will oblige, and deliver Per Aspera’s 2025 Wrapped, but succinctly, via text, without any over-the-top animations:

This year, Per Aspera published ~220,000 words. Despite leading very busy lives, on average, y’all opened the newsletter more than half of the time, and over 6,000 of you spent an extended period on the website reading our long-form Antimemos. From Issue 001 to 030 today, the community has grown more than 350%. The three most popular issues were 015, 017, & 018 - with September clearly being a gangbusters month.

2025 was great, but in the words of a startup founder who has just raised some money, we are just getting started. A huge thanks to all of you for coming on this journey with us; here’s to an even better 2026…

IN THIS WEEK’S EDITION:
🔊 Getting our airwaves in order
🛡️ Pax Silica, alliances, and economic clubs
🤝 Per Aspera friends in high, hard places

Forwarded this email? Subscribe to Per Aspera here.

In January 2022, all the major American airlines sent an urgent warning to the White House: if 5G networks switched on that week as planned, flights would be grounded across the country. Not because of weather, nor labor disputes, nor a new virus variant. But because two federal agencies had spent a decade using incompatible math to calculate how radio waves behave — and concerns weren’t elevated until $81B had already changed hands, and new wireless networks were deployed nationwide. Crisis was averted, but the underlying dysfunction remains.

Today's Antimemo is a deep dive into spectrum: the finite, non-renewable range of frequencies that carry everything from your phone signal to missile defense radar. Most Americans have no reason to ever think about the electromagnetic spectrum. It is invisible, silent, and effectively unnoticed — until something goes wrong. And things keep going wrong. In today’s essay, we:

The United States spent the last century building a theory of material power: steel, oil, and atoms. We have developed theories for semiconductors, for energy, and for critical minerals. We even have a theory of soft power. But we have yet to develop a coherent theory of invisible power: a theory of spectrum.

We treat it as a regulatory nuisance to be managed, or a commodity to be traded, rather than a strategic asset to be mobilized. The result: federal users sitting on 60% of the nation’s prime midband spectrum, technologies of the future starving for bandwidth, and a nation drifting toward isolating itself from global standards.

Too often, this gets framed as a choice between national security and economic competitiveness. This, dear reader, is a false choice.

The Grand Bargain we sketch out in today's piece (which really just rolls up what other smart folks and policy measures have proposed) would turn spectrum reallocation into a funding source for defense modernization, not a threat to it. A win-win, with better radars for the Navy, better allocations for economic growth and industrial users, a better footing for 5G & 6G, and crucially, a redirect away from Spectrum Island.

After reading today's Antimemo, you'll understand why it's past time we answer: do we govern the airwaves, or do they govern us?

The United States has formed a new club — Pax Silica — and given that it comes with a Latin name, Per Aspera is constitutionally obligated to approve. Last week, the State Department convened a one-day summit in Washington with eight allies — Japan, Korea, Singapore, the Netherlands, Israel, the UAE, the UK, and Australia — who collectively host critical nodes of the “cutting-edge AI supply chain.”

Pax Silica is three things at once: A) a club-based economic order, B) an affirmation of the value of AI, and C) a framework for coordinating trusted technology ecosystems, by supporting long-term offtake deals, expanding productive capacity across partner nations, and coordinating responses to overcapacity, dumping, and other coercive measures.

The scope of Pax Silica runs quite literally from sand (e.g, critical minerals & energy inputs) to the Dutch specialty — extreme ultraviolet (EUV) lithography — and beyond. What matters most is that this is a deliberate attempt to work across the full technology value chain, rather than pretend you can dominate one end while remaining utterly dependent at the other. Unofficially, it’s also a victory for system thinkers over specialists, and for anyone out there who’s been arguing we’ve ought to get miners and lithographers in the same room more often.

The New York Times editorial board has staked out its position on great-power competition, with the verdict being rendered in the picture above. In a long war, the NYT writes, victory is downstream of industrial capacity, and the U.S. no longer has enough of it by itself, with ~17% of global manufacturing, vs. China at 28% and growing. And, we’d note, those shares are typically measured in dollar value-added, which can overemphasize high-cost production and downplay sheer throughput. If you instead weight by physical output — units, displacement tonnage, ore mined and refined, industrial robot density — or by where the real value accrues in tightly integrated supply chains, or wartime, China’s advantage looks far larger and America’s smaller. The true ratio is debatable but the direction is not. The Times’ prescription is to treat allies as an extension of the industrial base, because “America + the club” can more readily counter China in key categories than by going it alone. But we have a bone to pick with the board’s implied framing of reindustrialization and free-world alliances as being at odds with one another. To this, we’d say: why not both?

All Hail The Queen!
Forbes put SpaceX President & COO Gwynne Shotwell at #20 on its 2025 list of the world’s most powerful women. Shotwell joined SpaceX in 2002 as its 11th employee and was immediately given an unglamorous mission: go sell a rocket that doesn’t exist yet to NASA and the Pentagon. The Gwynne Shotwell lore, for those of us who know her, is remarkably consistent: she is the executive who makes big, wild visions legible to serious buyers, then converts it into binding contracts, reliable schedules, operational capacity, and sustained flight rates. In 2008, after Falcon 1 reached orbit, she and Elon walked into NASA and secured the $1.6B ISS resupply contract that kept SpaceX solvent. She grew the Falcon manifest to hundreds of launches and $20B+ in contracted business. She speaks plainly, treats promises like liabilities, and builds trust through relentless execution in a line of work where almost equals failed. For 23 years she has been Elon’s operational complement, grasping his directional correctness about the future while ruthlessly calibrating his occasionally optimistic timelines with customer expectations and internal operations. On the eve of a $1.5T SpaceX public offering, this Forbes recognition belong to her — with one major quibble from our side: she should be way higher on the list!

Adapting rocket technology to defeat missiles mid-course
Long Wall has unveiled Cyclops, a surface-launched exoatmospheric interceptor built for mass-manufacturability (and scaling cost-effective magazine capacity). Midcourse interception remains brutally hard: extreme speeds, countermeasures, trajectories mostly outside Earth’s atmosphere, and a cost asymmetry that favors the attacker. Legacy interceptors are effective yet exquisite, expensive, and supply-constrained. Long Wall is attacking this by A) borrowing from industries with production rates orders of magnitude higher than defense, B) building its own liquid booster (RSX) and test infrastructure, and C) bringing 160,000+ sq. ft. of existing production facilities under one roof (including 3D printing, CNC, welding, assembly, hotfire testing, etc.) Formerly known as ABL Space Systems, Long Wall burned the boats and pivoted to missile defense in February, recognizing increasingly long odds in a saturated, cost-sensitive segment (small launch). Ten months later, it would seem the startup read the room correctly. For its next act, Long Wall says its job is now to “build and field [Cyclops] as fast as we can.”

Autonomy: Unsupervised, driverless Tesla seen on Austin streets (!!) // Waymo, meanwhile, is said to be raising $15B at a ~$100B valuation, with a $350M+ annual run rate Rivian lays out autonomy/AI strategy: E2E approach, custom (TSMC-built) silicon, 1x LiDAR , 5x radar, and 11x cameras…$2,500 one-time or $49.99/mo for L2 autonomy package // Luminar files for Chapter 11.

Robotics: UPS invests $120M in 400 truck-unloading robots // iRobot, the revolutionary company behind the Roomba, with 50M+ lifetime robotic units shipped, declares bankruptcy, will be absorbed by Chinese supplier — a cautionary tale, and cruel unintended consequence, of “antimonopolist” crusaders.

Automotive: Ford takes ~$20B writedown, among the biggest in history, in retreat from EVs to hybrids and ICE powertrains // VW ends vehicle production at 88-year-old Dresden factory, converts plant to AI, robotics, & silicon research hub // China approves two Level 3 autonomy cars from state-owned OEMs.

Orbit: Starfish & Impulse complete autonomous rendezvous and proximity mission in LEO // Overview Energy de-stealths with a unique take on space-based solar power, showcases power-beaming demo from a moving aircraft.

AI pushback: Democrat senators write to hyperscalers, seeking more information about AI datacenters’ role in rising electricity bills // Credit default swap volumes on AI-exposed tech players up 90% since September.

Reads: Viva la uranium renaissance // Is biology having its transformer moment? // How did the world change this year? Frontier of the Year 2025.

And, one for the road…File this one under ‘only in 2025’: SF woman gives birth in a Waymo en route to hospital. And this wasn’t even the first time a doctorless delivery in a driverless car has happened!