Key themes

AI in orbit: The compute moves to space

The satellite is no longer just a sensor — it is a compute node. AI running onboard (process in orbit, downlink the insight) and space-based data centers (orbital GPU clusters for LLM training) are racing to dominate. China already has 12 satellites live in its Three-Body Computing Constellation, with 2,800 more planned. The U.S. has no equivalent state program. Raw data downlinks hit a physical ceiling at ~10%; so in-orbit processing is not a roadmap item, it is the only scalable path.

Constellation for every nation — Sovereignty is the new anchor contract

44 countries have sovereign EO constellation plans. 5,770 EO satellites launch by 2034. Nations are no longer buying data — they are asserting independence. New commercial space operators are already the default suppliers for sovereign demand; incumbents that delay cede contracts permanently. 

So you want to build a spaceport — The infrastructure play

A viable spaceport can be brought from concept to first launch in roughly 2–3 years if treated as a disciplined infrastructure program. The critical early moves are picking strategically viable ground (inclination, safety envelope, logistics access) and locking an approvals path that derisks environmental, regulatory, and stakeholder objections. Parallel work streams on mission design, governance, facilities, and commissioning must be sequenced so that pads, range systems, and safety architecture converge on an executable concept of operations at first launch. 

Golden Dome — Money flows as architecture evolves

Congress appropriated ~$37B across FY2025–26 for missile defense related to the program as the architecture, industrial base and contracting sequence develops. Cost estimates range from $185B to $3.6 trillion, though this encompasses a range of capabilities. The real prize is battle management C2 — real-time AI integration of thousands of sensors and interceptors. 

Space is your offset strategy — The value chain is exportable

FMS hit $117.9B in CY2022 — space technology is dramatically underrepresented. ITAR and EAR are simultaneously the moat and the constraint: nations want US technology but resent the policy strings. Technology transfer and co-production (~30% of transactions) are converting buyer nations into competing defense exporters using US-origin IP. 

Human spaceflight — The Moon monetizes, Mars inspires

Legacy prime redundancy has collapsed. Vertically integrated commercial operators are now the national human spaceflight backbone. The cislunar propellant depot is the highest-NPV unbuilt asset in the solar system — every Mars mission and deep-space venture must pay for propellant. South pole water ice could cut Mars mission propellant cost from ~$5B to ~$500M per mission. Own the gas station, not the vehicle.

2026 Disruptive Predictions: P2P cargo

P2P suborbital freight. The battleground is the $3–7/kg air cargo layer. At ~50 full-vehicle reuses, rocket cost falls to $10–25/kg — below the cost-of-delay for semiconductors and biologics. The moat is the spaceport network, not the launch vehicle. 

2026 Disruptive Predictions: China's South Pole race

China and the Lunar South Pole. China was five years behind the U.S. in 2020. It is now months behind in several vectors. Chang'e-7 launches in 2026 for a south pole water ice survey; first crewed landing targeted for 2030 with all hardware prototyped. The ILRS Basic Station — nuclear-powered, 17 partner nations — is planned for 2035. The infrastructure window is 2026–2028.