Architecture is intent, poured into matter. It’s Ryan Duffy, Editor-in-Chief of Per Aspera.

Ludwig Mies van der Rohe — Bauhaus exile turned Chicago modernist — once said architecture reveals “the will of an epoch translated into space.” Walk a modern datacenter aisle and you feel the will of our artificially intelligent epoch: GPUs roaring, racks overheating, copper traces gasping, and a grid straining to feed an intelligence boom that doesn’t show any signs of slowing down.

The issue? Electrons, the heroes of our industrial internet, are finally running out of runway. So where do we go when the substrate buckles? This week’s antimemo answers with light. Photonics promises racks that run cold, clusters that breathe, and full-stack infrastructure built on light. We’re calling it “Architecture Movement: Photonics” and we can’t wait to hear what you think.

P.S. Know a smart friend who should be reading Per Aspera? Forward them this email. And if you’re that smart friend, sign up here.

“Architecture Movement: Photonics.” For more than a century electrons carried our ambitions: spinning motors, lighting screens, and sluicing data through copper veins. From Morse’s telegraph to Bell’s telephone to the transistor revolution, we refined the art (and science) of flipping bits with electrons. Today, supercomputers like xAI’s 100,000+ GPU “Colossus” stand as the skyscraper equivalent of the electronic era, the pinnacle of an architecture built on charge.

But AI is rewriting the demand curve, with scale, speed, and an endless appetite for compute that laughs at yesterday’s power budgets. Racks cook, latency spikes, utilities say no más. We’re close to hitting the load-bearing limit of copper and electrons.

So…what comes next? A material shift. Every architectural leap starts there. Today, we go searching through the periodic zoo of particles and plant our flag on the humble photon.

Enter photonics. Data rides on light, the cosmic speed limit. Photons shed almost no heat and thread terabits through glass slimmer than a cactus spine. After decades of quiet progress, the technology now promises to outclass copper at every hopper. Pair photons with silicon logic and the machine can finally breathe: electrons compute, photons move, each in its natural lane.

*To be sure, photons won’t arrive as a Hollywood montage of white lab coats high-fiving. They show up as sub-micron alignment nightmares, toolchains missing the “compile” button, and venture decks that read like someone asking you to build the next TSMC. Exactly the kind of hard frontier Per Aspera was built to scout.

Whether you run a cloud region or just run ChatGPT in your browser, the substrate shift will shape your world. It’s equally critical for those of you underwriting the next hyperscaler park; for engineers at TSMC, ASML, Samsung, Intel, and NVIDIA thinking about the fabs of the future; for PE funds eyeing infrastructure bets; for grid operators clearing gigawatt queues; and for policymakers tasked with hardening national resilience.

(Also note Beijing’s coordinated push, from Made in China 2025 to the 14th Five-Year Plan, to secure leadership in photonics and stand up labs dedicated to light-based technologies.)

If you care about performance, power, profit, and progress in the Age of AI, you’ll care about this. So, pour a cup of coffee, open the antimemo, and meet the particle that could keep the lights on — by letting a little light in.

Hello, Dan Goldin here. This is for you investors in the crowd.

For those who don’t know me, I was the 9th NASA Administrator and prior to that, GM of Space & Technology at TRW, which was acquired by Northrop Grumman after the 1993 Last Supper.

I want to say something that few others will. I believe the future of humanity does not have enough risk-taking investors in the room. What I mean is that the techie world is touting an incredibly advanced, deeply technological future.

But the private capital side does not seem to be matching that vision.

I get it. I now advise private equity and venture capitalists. I know how hard it is to part with capital for risk. But I’ve lived through enough tech revolutions to know what happens when we don’t invest.

And, more importantly, I’ve seen the magic that happens when we do.

I was there for the mainframe. Then, when I got pulled into making bombs and bullets, I went on to DARPANET. After that, the internet arrived. It was a total rewiring of how the world worked, and it only happened because people bet on the infrastructure, tools, and ideas that seemed too capital intensive (at the time). Because they understood the — extreme — implications of connecting humans across the universe.

Now we’re heading into the next frontier: intelligence. AI is going to need new infrastructure. For all the work that they’ve done for us, electrons simply can’t carry us much further. They’re tapping out — too hot, too slow, too power-hungry.

We need to move. We need to fund new substrates. New bets. So, I’ll ask you the question that matters:

Are you going to be someone who funds the future of humanity?

Or are you going to be some who chases sub-20% IRRs, because that’s better than T-bills?

BTW…a lot of you have asked me how to get involved. Here’s my one ask: Forward this email to someone who cares about the Renaissance we’re in. That’s how this movement grows, from one enlightened person to another!

Historically, precision navigation has relied on fiber optic gyroscopes (FOGs) — bulky, costly systems used in ICBMs, fighter jets, and submarines. ANELLO Photonics collapses that complexity into a silicon chip. Their breakthrough, the SiPhOG™ (Silicon Photonics Optical Gyroscope), miniaturizes the physics of FOGs using integrated photonics, resulting in a coin-sized device that delivers the precision of a fiber-optic gyroscope at a fraction of the size and cost. Unlike MEMS gyros found in smartphones — which are small but imprecise — SiPhOG offers long-duration inertial guidance without GPS, enabling autonomy in drones, robots, autonomous vehicles, underground mining vehicles, and even maritime vehicles. By leveraging silicon photonics — a field ANELLO’s team pioneered at Intel — the SiPhOG is immune to vibration, temperature shifts, and electromagnetic interference, while being scalable for mass production. The SiPhOG fills the precision gap between low-end MEMS and high-end fiber optic gyroscopes — opening up an entirely new layer of autonomy infrastructure. In an era where GPS signals are routinely jammed or spoofed, ANELLO’s chip-scale solution delivers resilient, high-fidelity navigation across land, air, and sea.

This Situation Report provides our weekly pulse check on hard pursuits, industrial developments, and deep tech — a curated snapshot of what matters, from test stands to launch pads, from lab benches to factory floors. Real signals, no BS.

🛰 A (**checks notes**) $146M Series B. Muon Space has announced a ~$90M Series B-1, bringing the round’s total to an eye-popping $146M. The Mountain View startup isn’t just slinging and shipping satellites; it’s taking key manufacturing principles to an altitude of 500 km. CEO Jonny Dyer calls Muon “the world’s first automated, high-mix, high-volume constellation manufacturing system.” The fresh capital, a buyout of Starlight Engines (maker of compact, zinc-fueled Hall thrusters), and an expansion into a new 130,000-sq-ft San Jose facility will help the company service the $100M+ in new contracts signed in 2024 alone. After a bit of a SPAC-induced fundraising lull, we’ve seen a slew of strong rounds from growth-stage space players. Are we so back?

🔩 Steel in the Stack. Deep tech held relatively steady in Y Combinator’s latest Demo Day, with ~11% of the cohort in the “hard tech” category (vs. a historical average of around half that). An impressive feat, considering that AI agents and their ilk have inhaled nearly every molecule of startup oxygen mindshare in SF. Meanwhile, CNBC’s 2025 Disruptor 50 leans even harder into the world of atoms, with Anduril taking the #1 spot and roughly a third of the list building for the physical world, including defense tech (Shield AI, Saronic), autonomous fleets (Waabi, Einride), humanoids (Apptronik), precision ag (Carbon Robotics), and drone logistics (Zipline). Finally, the sharpest pulse comes from the new Thiel Fellows, who are working on concepts like programmable matter, a planetary defense system, non-invasive neurostimulators, and more — shipping prototypes before most peers finish sophomore year. More of this, please!

💾 ATC running on nostalgia. Anyone who’s tried to fly through Newark, NJ recently knows firsthand that our air traffic control system is struggling. U.S. ATC centers still run Windows 95, floppy disks, and paper strips (not to mention, radar gear ready for the Smithsonian). The FAA’s new RFI aims to overhaul 51 legacy systems, but a 24/7 network must swap like a heart transplant, not a routine IT refresh. As some have noted, ancient code may ironically be more insulated from modern cyberattacks, yet recent NOTAM and radar issues underscore the rising risks of inaction. The FAA claims bipartisan backing and a four-year runway to swap disks for data — but with no price tag on the table and a history of aborted takeoffs, tough questions remain around whether the U.S. can land this upgrade. Stay tuned…an antimemo is in the works on the bold plan to rewire America’s airspace nervous system.

In Case You Missed It, last week’s antimemo took aim at the water crisis in the American Southwest, and proposed a bold alternative to defeatism and decline.

While much of the American Southwest clings to rationing and half-measures, The End of Thirst Traps argues for something bigger: a modern megaproject that trades sun and seawater for sustained growth. A solar-powered desalination and distribution system stretching from the Gulf of California to Phoenix, Vegas, and the Salton Sea. This isn’t a sermon on scarcity — it’s a blueprint for abundance and engineering at scale.