Good morning, Dan Goldin here. On Sunday, I received a message from Low Earth Orbit. After 60+ years in the biz, getting a personal note from someone living in outer space still feels electric!

It was from Colonel Michael Fincke onboard the ISS right now (everyone say: Hello, Col. Fincke!) — a pilot for SpaceX’s Crew-11 mission, prior Space Station Commander, and one of the last astronauts trained in both the Shuttle and Soyuz eras. A true American spacefarer.

The ISS was built in an era of doubt, when much of America questioned whether space was still worth reaching for. The fact that astronauts are living there today — and phoning home from orbit! — is proof of what we can achieve when we commit to hard things. And in Michael Fincke’s journey, from Shuttle/Soyuz to SpaceX, we see how far we’ve come: from government-built outpost to a new public/private era.

It’s the 25th anniversary of the ISS. I’ve written at length about ISS history, so today I want to talk future. The question before us: What must we do to lead a spacefaring civilization?

IN THIS WEEK’S EDITION
🧑‍🚀 Seven Thoughts for America's Next Space Age
🛰️ Much ado about datacenters in orbit
💸 The Renaissance’s capital stack
🤝 Friends in high, hard places

Forwarded this? Subscribe for more.

Twenty-five years ago this week, humanity established a permanent foothold in the cosmos. On Nov. 2, 2000, at the turn of the millennium, the International Space Station received its first crew, beginning an unbroken chain of human presence beyond Earth that continues this moment, as you read these words.

I remember testifying before Congress, making promises that seemed impossibly ambitious:

The budget hawks descended on us, and critics declared it a fantasy. In June 1993, the hopes and dreams of the entire American human spaceflight program survived by a single vote: 216-215. Congressman John Lewis cast the deciding ballot. What a moment in history. I only wish Congressman Lewis was still with us today to share this joy. 

The ISS has lived up to every promise I made to Congress. It has proven to be one of the most complex, safe, and enduring engineering programs in the history of the planet. And the scientific promise of the outpost has delivered, too:

What made it possible were thousands of technical details that nobody sees: closed-loop life support, CO₂ scrubbing, ammonia cooling loops, redundancy in power supply, ways to swap out air purification units or repair a sealed hatch in real time. The ISS was a forced experiment in reliability: we studied failures, engineered fixes, and iterated on hardware and process until the lessons stuck. Crews learned to carry out real science, maintain systems, and improvise solutions.

Here's what I've come to believe, and it may shock you coming from the man who fought to build it: The ISS needs to come down. Not because it failed. Quite the contrary: it succeeded beyond our wildest dreams! But that very success has become our trap. We're still locked to LEO, comfortable in the transit lanes built by Apollo and the shuttle, dependent on endless resupply missions, and beholden to Mission Control for seemingly every little decision.

To go further, we need to break genuinely hard new barriers. Every dollar extending the ISS’s useful life has an opportunity cost, as it’s not being spent on breakthrough propulsion, closed-loop life support for deep space, or the infrastructure needed to reach the Moon and Mars.

If we’re serious about escaping Earth orbit and truly becoming a spacefaring civilization, here’s what it will take.

001 // Cut the cord to mission control. At its farthest, Mars is 21-24 minutes away, one way, at light speed. By the time Houston learns you’re dying, you’re already dead. Crews need reliable, physics-based AI to diagnose and fix failures in seconds, and the autonomy to act without Earth’s permission. Comms should carry science and strategy, not survival instructions!

002 // Make biology our life support system. Mechanical recycling has plateaued at 50% efficiency, despite decades of optimization on the ISS. Earth achieves 100% through biology. We need photosynthesis, algae, fungi — systems that turn waste into food, CO₂ into oxygen, urine into water. The difference between 50% and 95% is the difference between endless resupply runs and permanence.

003 // Embrace the robots. A suited astronaut works seven hours before exhaustion. A spacefaring humanoid robot could work continuously for months. We need machines that share our tools and workspaces, that can swap out a faulty component or mine regolith using the same equipment humans would use, but without prebreathe protocols, suit checks, radiation limits, or fatigue. One operator commanding ten humanoid robots could accomplish more than a dozen astronauts for a fraction of the cost and risk.

004 // Generate power at industrial scale. Solar works in Earth orbit. But what about a lunar base with 336-hour nights? We need multi-MW nuclear systems running autonomously for decades. Power enables everything: ore processing, propellant production, manufacturing. Every kilowatt we can't generate is another supply mission we can't eliminate.

005 // Graduate from chemical propulsion. We've spent a century perfecting chemical rockets, squeezing marginal gains from the rocket equation. Nine months to Mars inflicts bone loss, muscle atrophy, radiation, and psychological strain that exercise can’t fully counter. Nuclear thermal propulsion doubles efficiency. Nuclear electric improves it tenfold. Fusion, when we achieve it, could reduce Mars transit to weeks. My NASA team proved scramjet feasibility at Mach 9.68 in 2004; that record endures because we chose incremental rockets over revolutionary hypersonic aircraft development.

006 // Use what's already there. Lunar regolith is 40% oxygen. Mars offers CO₂ for fuel when mixed with hydrogen. Water ice becomes propellant or shielding. Yet we haul everything from Earth at thousands of dollars per kilogram. We need the full chain: prospecting, extraction, refining, and finishing. Industrial-scale excavators for abrasive regolith, smelters that work in vacuum, and 3D printers that build habitats from local materials.

007 // Let government and markets do what they do best. NASA should push for breakthrough physics and the search for life: the fundamental questions only government will fund. For everything else (transport, LEO stations, mining, fabrication), industry is ready to take the baton. (For the ISS specifically: retire it, and redirect the $3B/year budget toward purchasing commercial station services.)

America needs NASA, now more than ever. Not as a trucking company to LEO, but as the pathfinder to the impossible. NASA builds the instruments that answer fundamental questions about the origin and destiny of our universe. Its calling is to push for breakthroughs in propulsion, power, life support, and flagship science missions — technologies that markets won’t touch today because they’re too hard, or too far from profit.

Twenty-five years have proven the concept: humans can survive in space for months. But survival ≠ settlement. The ISS united the world, taught us how to live in orbit, and opened the door to commercialization. Mission accomplished. Now the world is ready for the next chapter: permanent outposts on the Moon, resources from near-Earth asteroids, boots on Mars, and beyond.

The Station was always meant to be a springboard, not a destination. Now it’s time for us to jump.

On Sunday, Starcloud achieved a first, deploying an NVIDIA H100 into low Earth orbit (the H100 is a datacenter-class GPU). Starcloud is a “datacenters in space” startup that graduated from Y Combinator last summer. The startup, its NVIDIA partnership, and the larger concept of orbital datacenters have been a lightning rod among engineers, founders, and physicists online in recent weeks.

At Per Aspera, we believe the only way to honor this wave of ambition is to match it with clarity. Most space-based compute conversations still live in the realm of marketing and hype. Ours lives in thermal budgets, power cycles, radiation hardening, orbital dynamics, and structural deployment. If you want to cut through the noise of space-based datacenters, start here.

Apollo Global Management, Inc. and 8VC have announced a strategic partnership that will see the two deploy “several billion dollars” to fund advanced manufacturing, space, energy, life sciences, and natural resources startups.

This is good news! We’ve seen a massive funding gap for companies that don’t pattern-match for consensus megafund investors but nonetheless need sophisticated capital structures and domain expertise. Models like this may help fill that gap, providing a capital stack that makes the Renaissance viable at scale.

Purdue University’s Joe Jewell appeared before the Senate Armed Services Committee for his confirmation hearing as Assistant Secretary of War for Science and Technology.

In his testimony, Jewell reflected on a family legacy of service dating back to the Revolutionary War. He connected that heritage to his mission of advancing America’s defense leadership through science, and called for the U.S. to close the gap between discovery and deployment, “shortening the timeline from laboratory breakthroughs to battlefield capabilities.”

For the uninitiated, Purdue has been leading an education revolution to rebuild American capacity — with even Citadel’s Ken Griffin calling it out as one of the “real success stories in American higher education.” Go Joe, go PARI, go Purdue!

👉 ICYMI, we partnered with PARI CEO Mark Lewis to break down what it will take to lead in hypersonics. Read the writeup here.

Space: To The Moon and Beyond: a strategically timed update from SpaceX on HLS, Artemis 3, a “simplified” architecture, and 49 milestones the team has cleared // FCC floats ‘licensing assembly line’ to streamline satellite approvals // China’s ZQ-3 could eventually outperform Falcon 9, likely long after upgraded Starships are routinely flying // Apex Space, an up-and-coming, well-capitalized LA-based satellite bus developer, aims to build and sell (but first, demonstrate) space-based interceptors.

Chips & Compute: U.S. DoE and AMD form $1B partnership to build 2 supercomputers: Lux & Discovery // Blackwell chips are now in full production in Arizona // Anthropic will buy up to 1M TPUs from Google, and is also running Claude on 1M of Amazon’s Trainium 2 accelerators, as labs diversify away from sole reliance on NVIDIA accelerators // While we’re here: NVIDIA held GTC, aka the Super Bowl of AI, in DC last week (a first), where it announced new/expanded partnerships with Nokia, Palantir, CrowdStrike, Oracle, HPE, DoE, Uber, Eli Lilly, Dell, and more. The vibes:

Robotics: 1X launches Neo, a $20K humanoid housekeeper robot: teleoperated for many tasks, available now for preorder, deliveries expected to start next year // WSJ takes on the “hands problem” holding back humanoids, which is something we’ve harped on: dexterity is the long pole of the tent // Elon, on Tesla’s recent earnings call, said: "It’s an incredibly difficult thing…to create a hand that is as dexterous and capable of the human hand.” // iRobot, trailblazing pioneer of an early robotics category, has going concerns, as it’s been unable to find a buyer (recall that the FTC blocked Amazon’s $1.4B takeover of the Roomba maker….) // And, for the road, here’s a fascinating profile of the Filipino teleoperators of Japan’s convenience store robots.

Etc… Have batteries helped reduce rolling blackouts in California? // Google’s quantum/AI group published a blog post and Nature paper on its “first-ever verifiable quantum advantage.” // Could the AI datacenter boom be masking a lagging manufacturing revival? // Enedym, a Canadian startup developing rare-earth-free switched reluctance motors, landed a strategic investment from Honda.