Everyone’s hyping space-based data centers — no one’s doing the physics. It’s Ryan Duffy, Editor-in-Chief of Per Aspera.

There’s a lot of talk about space-based data centers — ambitious decks, futuristic renders, and endless optimism about offloading compute to orbit. But beneath the glossy concepts lies a missing conversation: the economics. And in this anti-memo, we’re doing what almost no one else has — digging into the actual financial gravity of space-based compute. Not the sci-fi pitch, but the real unit economics mapped to engineering constraints: launch costs, thermal efficiencies, power sourcing, hardware redundancy, latency penalties, and more. We break down where the economics collapse, where they might pencil out, and what levers have to move for this idea to become viable. The dream is bold. But the spreadsheet is brutal. We’re calling it “Realities of Space-Based Compute”.

Welcome to Per Aspera's second antimemo. In these, we dive deep into crucial technological frontiers that will shape our future. This week, we explore a topic that sits at the intersection of advanced computing, space, and geopolitics.

“Realities of Space-Based Compute.” Not too long ago, our corner of the internet lit up with a news story — Eric Schmidt, the former CEO of Google, had acquired rocket manufacturer Relativity Space. The move seemed puzzling until a reporter connected the dots: when asked if he bought an aspiring heavy-lift launcher to put data centers in orbit, Schmidt's one-word reply— "Yes" —was confirmation enough.

Then, last week, a Chinese rocket delivered the first 12 satellites of what Beijing calls their "Three-Body Computing Constellation." It was the opening volley in a planned 2,800-satellite network designed to deliver an eye-popping 1,000 POPS of distributed computing power by the 2030s.

From Palo Alto to the Pentagon, there is brimming fascination with the concept of orbital data centers. This isn't just speculative futurism. The technology is moving from slideware to flight hardware, and space is now on the short list of places serious people look for the next jump in compute capacity.

THE POWER PARADOX

The timing isn't coincidental. On Earth, our computational ambitions are increasingly constrained by physical infrastructure. Tech giants struggle to power their AI ambitions and terrestrial infrastructure could hit hard limits.

Resource demands are manifesting in more and more places: from Silicon Valley power queues to Phoenix's drought-exacerbating cooling requirements to Northern Virginia's residential-to-industrial rezoning battles. By 2030, AI could consume nearly a tenth of U.S. electricity generation—requiring 60 nuclear plants’ worth of new capacity.

WHAT IF WE LOOK SKYWARD?

Space beckons with a seductive promise: “infinite” solar power, perfect vacuum cooling, and not a single NIMBY in sight. The perfect end-run around Earthly bottlenecks. The vision is elegant—but the physics are brutal.

At Per Aspera, our antimemos are comprehensive technical analyses that go beyond typical industry reports — we’ve spent months researching this topic, and produced a 14,000+ word beast of a deep dive that reveals:

1. How space inverts traditional computing economics

2. Five applications where orbital compute could be worth the premium

3. The engineering trade space that must be navigated

4. A three-stage roadmap from today's kilowatt demonstrators to future megawatt-class systems to scaled platforms

5. A reality-based playbook for investors and builders to underwrite this emerging frontier

BEYOND HYPE CYCLES

The economics only work at precise intersections of competing constraints. Success requires mastering both physics and finance—pairing engineering breakthroughs with applications that justify the premium of putting silicon in orbit.

This is what Per Aspera exists to provide: to cut through the froth. This antimemo will be your technical compass, helping you navigate an arena destined to reshape AI and computing infrastructure discussions for decades to come.

OFF-CUFF

A note of thanks from Dan Goldin, Ryan Duffy, Jeff Crusey, and Joy Shin — the four creators of Per Aspera.

We’re at T+7 days, and what a week it's been since our launch! A sincere thank you to the thousands who've joined us. Your response has been extraordinary—signup rates tripled our projections, and your feedback has been both encouraging and thought-provoking.

Our inaugural antimemo — "Pray Not for Easy Lives, But to Be Stronger" — struck a chord with y’all, particularly among those deep in the hard pursuits we champion. Defense founders, aerospace engineers, materials scientists, and “Made in the USA” leaders have reached out with detailed feedback (you know who you are – thank you). Your notes affirm that we're building something vital.

This week's deep dive into space-based computing exemplifies exactly the "hard pursuit" ethos we discussed—where physics constraints, engineering challenges, and strategic implications collide. While our first piece explained the "why" of tackling difficult problems, this antimemo tackles the "how" with uncompromising analysis.

Several of you asked about our publishing cadence; we're committed to quality over quantity, with antimemos and newsletter editions dropping each Monday. Others inquired about topic selection—we focus on areas where conventional wisdom misaligns with physical reality, or where strategic interests and technical capabilities intersect in underexplored ways.

Housekeeping note: We'll be off next Monday for Memorial Day, returning the following week with our next edition.

An ask of you: What domains should we add to our pipeline? Small modular reactor economics? Advanced materials for hypersonics? The future of precision manufacturing? Drop us a line by replying to this email.

This is a living experiment. Building a high-signal publication isn't just about pushing content—it's about fostering conversations that matter. We're here for the long haul, championing those of you doing the difficult, necessary work.

— Dan, Ryan, Jeff, & Joy

P.S. If you were forwarded this email, subscribe to Per Aspera by clicking below.

For decades, compute infrastructure has remained Earth-bound—limited by land availability, grid dependence, cooling demands, and bandwidth bottlenecks. Sophia Space, a portfolio company of Mandala Space Ventures, was purpose-built to challenge that constraint. Backed by Mandala’s deep expertise in next-gen space systems and led by Dr. Leon Alkalai, Rob DeMillo, and Brian Monnin, Sophia delivers a modular, space-native compute platform that enables radiation-tolerant, solar-powered, passively cooled AI processing in orbit. Each TILE™ functions as part of a resilient mesh, with onboard acceleration and optical interlinks delivering real-time inference without relying on terrestrial infrastructure. This architecture makes orbital cloud computing viable at scale—supporting ISR pre-processing, closed-loop autonomy, and fault-tolerant AI across defense, civil, and commercial applications. Strategically, Sophia marks a foundational shift in U.S. digital infrastructure: sovereign, upgradeable, and distributed orbital compute—critical for scaling ISR beyond spectrum constraints, ensuring continuity in denied environments, and securing the future of autonomous space systems.

OTHER NEWS

This section 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.

🚀 Breaking The Detonation Barrier. Huge round of applause for our friends at Venus Aerospace for notching the first U.S. flight of a rotating detonation rocket engine (images courtesy of Venus Aerospace, above). Their 2,000-pound-thrust test at Spaceport America moves this tech from theory to hardware-a real milestone for high-efficiency propulsion. Rotating detonation engines have been a white whale for decades; now, Venus is not just testing but building out the infrastructure to keep pushing. This is how you move the needle on American aerospace.

🛡️ Aluminum Moves West. Oklahoma has signed a $4 billion deal with Emirates Global Aluminum to build the first new U.S. aluminum smelter in 45 years. The plant’s 600,000-ton annual output will chip away at U.S. reliance on imported aluminum—a material critical to aerospace and energy.

🥷🏻 Online-to-Offline Attacks. Nucor, North America’s top steelmaker, took some production offline last week following what they clinically termed a "cybersecurity incident." As more physical infrastructure gets wired into sprawling IT networks, the attack surface grows and the risks multiply. No attribution, no ransom note, just a clear signal: the line between cyber and real-world disruption is gone, and most U.S. industrial systems are still playing catch-up

🏴‍☠️ Making Space Manufacturing Boring. Varda Space Industries successfully reentered its W-3 capsule at the Koonibba Test Range in South Australia, marking their third mission just 11 weeks after the last. Varda is rapidly establishing a cadence of autonomous, high-hypersonic reentries, building critical infrastructure for both space manufacturing and hypersonic testing. What's remarkable is how unremarkable it felt: no breathless live coverage. As Varda quipped after the mission: “Is this boring yet?” Like SpaceX launches—once historical events, now routine operations—Varda's manufacturing missions are transitioning from novelty to repeatability. This pattern can be found across all successful deep tech: the Wright brothers' first flight was global news; today's 100,000+ commercial flights daily merit zero headlines. The ultimate metric of industrialization isn't excitement but normalization.