There you have it! The Space Force just published its 15-year blueprint — 30,000 satellites by 20…
April 18, 2026 · 0 likes · 0 comments
China Threat Defense Workforce AI
There you have it! The Space Force just published its 15-year blueprint — 30,000 satellites by 2040, AI and autonomy fused into every constellation, ground station, and orbital proximity operation. Sensing, deciding, and acting at MACHINE SPEED.
The vision is right. Finally.
Gen. Saltzman said it out loud at Space Symposium: "We simply cannot wait for 'perfect' any longer." Minimum Viable Capabilities. Proliferated LEO. A Space Data Network that orchestrates itself. A second NavWar squadron. Gen 4 PNT because today's GPS was never designed for a contested domain.
Every word of that document is correct.
Now read the next sentence carefully.
I was the first Chief Software Officer of the U.S. Space Force. I wrote the software strategy. I sat in the rooms where these plans get born — and then quietly strangled by the same acquisition machine that took 15 years to ship a software upgrade that STILL doesn't work in combat.
Plans don't fly. Software does.
A 160-page vision document is not a satellite. It is not an AI model. It is not a kill chain. It is a PDF. And the Pentagon has warehouses full of brilliant PDFs that never reached orbit.
Let me tell you what 30,000 satellites actually means.
It means 30,000 attack surfaces. Every bus, every payload, every ground link, every firmware update — a door. China will spend the next 15 years looking for the weakest one. And they launch weekly while we write white papers.
It means "machine speed" is a BLUFF unless the AI actually exists, runs on classified networks, is trusted by operators, and ships faster than an adversary can iterate. Right now the Pentagon is BANNING the most safety-aligned frontier AI labs in the world while pretending it will have autonomous orbital warfare by 2040. You cannot do both.
It means the same acquisition machine that can't field a GPS receiver in under a decade will be asked to field an AI-orchestrated constellation of tens of thousands of nodes. Same primes. Same cost-plus contracts. Same Valley of Death.
This plan will be the rebirth of American spacepower — or the most expensive PowerPoint deck in U.S. history. No middle outcome.
Saltzman knows it. That's why he wrote "speed is now a strategic requirement, not just an efficiency goal." Read that twice.
The execution — not the plan — is the war.
Commercial AI. COTS hardware. Proliferated LEO. Iterative software. Fire the primes that won't deliver. Pay the ones that will. Let the Space Force Guardians ACTUALLY compete with Chinese engineers instead of compete with each other for conference slides.
I'm writing about exactly this in my book, "Replacement" — out end of July. How America either becomes a country that BUILDS at machine speed, or gets replaced by one that does.
30,000 satellites is not a goal. It is a DEADLINE.
Execute or lose the domain.
What are your thoughts?
The vision is right. Finally.
Gen. Saltzman said it out loud at Space Symposium: "We simply cannot wait for 'perfect' any longer." Minimum Viable Capabilities. Proliferated LEO. A Space Data Network that orchestrates itself. A second NavWar squadron. Gen 4 PNT because today's GPS was never designed for a contested domain.
Every word of that document is correct.
Now read the next sentence carefully.
I was the first Chief Software Officer of the U.S. Space Force. I wrote the software strategy. I sat in the rooms where these plans get born — and then quietly strangled by the same acquisition machine that took 15 years to ship a software upgrade that STILL doesn't work in combat.
Plans don't fly. Software does.
A 160-page vision document is not a satellite. It is not an AI model. It is not a kill chain. It is a PDF. And the Pentagon has warehouses full of brilliant PDFs that never reached orbit.
Let me tell you what 30,000 satellites actually means.
It means 30,000 attack surfaces. Every bus, every payload, every ground link, every firmware update — a door. China will spend the next 15 years looking for the weakest one. And they launch weekly while we write white papers.
It means "machine speed" is a BLUFF unless the AI actually exists, runs on classified networks, is trusted by operators, and ships faster than an adversary can iterate. Right now the Pentagon is BANNING the most safety-aligned frontier AI labs in the world while pretending it will have autonomous orbital warfare by 2040. You cannot do both.
It means the same acquisition machine that can't field a GPS receiver in under a decade will be asked to field an AI-orchestrated constellation of tens of thousands of nodes. Same primes. Same cost-plus contracts. Same Valley of Death.
This plan will be the rebirth of American spacepower — or the most expensive PowerPoint deck in U.S. history. No middle outcome.
Saltzman knows it. That's why he wrote "speed is now a strategic requirement, not just an efficiency goal." Read that twice.
The execution — not the plan — is the war.
Commercial AI. COTS hardware. Proliferated LEO. Iterative software. Fire the primes that won't deliver. Pay the ones that will. Let the Space Force Guardians ACTUALLY compete with Chinese engineers instead of compete with each other for conference slides.
I'm writing about exactly this in my book, "Replacement" — out end of July. How America either becomes a country that BUILDS at machine speed, or gets replaced by one that does.
30,000 satellites is not a goal. It is a DEADLINE.
Execute or lose the domain.
What are your thoughts?