China's Commercial Space Industry Replicating Solar & Lithium Battery Cost Reduction Curves, Approaching Commercialization Inflection Point
Miles Bennett
UBS says China's commercial space sector is nearing a commercialization inflection point — launch costs have fallen from $10,000–15,000/kg in 2019 to roughly $4,000/kg, and could drop to $900–1,900/kg by 2030 if learning rates hold, mirroring the scale-driven cost curves of solar panels and lithium batteries.
Launch costs slashed in six years — how?
The logic is classic manufacturing: scale up, costs come down. In plain terms = the more you build and launch, the cheaper each launch gets — solar and batteries followed the same playbook.
Regression analysis confirms the parallel: China's solar module learning rate is roughly 34.9%, lithium batteries about 26.2%, and commercial space sits in the 20%–35% range — the curve shapes closely match.
By 2025, cumulative commercial launches reached about 95. If that approaches 1,000 by 2030, cost could drop from today's ~$4,000/kg to $900–1,900/kg.
Reusable rockets — where is the single biggest cost lever?
The most expensive part of a rocket is its first-stage engine. This means → if the first stage can be recovered and reused, the cost curve bends sharply.
LandSpace estimates that reusing the first stage five times cuts per-launch cost by up to 45%. The company completed a milestone launch-and-recovery test in Q4 2025; the Long March 12A passed a similar test in the same period.
Two more milestones to watch in H2 2026: Galactic Energy's Ceres-1 and iSpace's Hyperbola-3, both planning first flights with recovery attempts. Reusability in China, however, remains in the validation stage — batch commercialization is still some distance away.
Supply-chain commercialization — the overlooked cost path?
Space Pioneer's data shows roughly 95% of its rocket components can be sourced from automotive, aviation, and machinery suppliers. In plain terms = rockets are no longer "specialty manufacturing" — most parts come from the same supply chains that build cars and planes.
But about 30% of components still come from the legacy state-owned aerospace system. This means → every percentage point that share drops opens another notch of cost room through market-rate procurement.
From 1,333 to 50,000 satellites — what are the three hurdles?
As of Q1 2026, China has roughly 1,333 satellites in orbit (remote sensing 46%, communications 35%). The GW and Qianfan constellations target 50,000 — about 40× the current fleet.
Three hurdles stand in the way: ① reusable rockets are not yet commercially proven; ② current launch capacity cannot support the 10,000+ tonnes of deployment needed; ③ satellite power, thermal management, and payload technologies are still maturing.
There is also a hidden clock: ITU rules require 100% deployment within 15 years of spectrum filing, or the filing lapses. This means → GW and Qianfan combined must deploy over 15,000 satellites within the next decade; Qianfan Phase 1 (1,296 satellites) is slated to begin in 2027.
Space-based computing — sounds like sci-fi, so what can actually land soon?
The rationale: AI is pushing ground-based compute into bottlenecks. Satellites in dawn-dusk sun-synchronous orbit — a path that keeps them in near-constant sunlight — can harvest solar power almost continuously, benefit from radiative cooling in space, and face no ground-level permitting constraints.
But for orbital data centers to truly replace terrestrial ones, today's launch cost of ~$3,000/kg and space-grade solar panel cost of ~$10,000/kW (P-type HJT) both need to fall roughly 80% to reach grid-power parity. This reflects how far space computing remains from full commercialization.
The nearer-term payoff is on-orbit data processing — satellite imagery and SAR data (synthetic-aperture radar — high-resolution ground imaging via radar) processed directly in orbit, cutting the need to downlink to ground stations. In May 2025, China launched its first batch of 12 computing satellites, forming the prototype of the "Three-Body Computing Constellation."
On-orbit servicing — why can extending a satellite's life reshape the entire business model?
LEO satellites typically have a design life of 5–7 years. In plain terms = for computing satellites packed with processing hardware, a 7-year lifespan makes the investment-return math very hard to close.
Emposat recently completed an on-orbit test using a robotic arm for fuel resupply in low-Earth orbit — China's first commercial test satellite equipped with a flexible robotic arm. This means → if on-orbit life can be meaningfully extended, the business model for computing satellites changes at its root.
Whether Qianfan Phase 1 launches on schedule in 2027 will be the first real test of whether China's commercial space supply chain can handle genuine order-volume pressure.
Content is for reference only, not financial advice.