To Boldly Go: The Case for Space Datacenters
SemiAnalysis Twitter · SemiAnalysis (@SemiAnalysis_) · 2026-06-03
SemiAnalysis publishes a detailed total-cost-of-ownership analysis arguing that space-based datacenters could reach terrestrial cost parity in the late 2030s, but only after chip manufacturing constraints are resolved and major reductions achieved in launch, radiator, and solar array costs.
Appears in
Extraction
Topics: space-datacentersorbital-computeai-infrastructuresemiconductor-supplydatacenter-economics
Claims
- Deploying orbital compute using current technology costs several times more than equivalent terrestrial compute.
- Space-Earth datacenter cost parity is achievable in the late 2030s under favorable engineering and cost-scaling conditions.
- Chip manufacturing capacity—specifically TSMC N3 wafers and HBM supply—is currently the primary global constraint on AI compute expansion, ahead of datacenter power or space capacity.
- Common pro-space-datacenter arguments such as free solar, free cooling, and low latency are superficially appealing but fall apart under rigorous engineering analysis.
- SpaceX's S-1 filing targets 100 gigawatts of compute launched to space annually as a long-term goal, framing space compute as central to its IPO narrative.
Key quotes
While we think that it is possible that space datacenters could scale one day, deploying orbital compute using today's technology currently costs several times more than deploying terrestrial compute.
Space-based datacenters make sense in the world where AI demand well exceeds all of the four layers of terrestrial datacenter supply.
Five years from now, my prediction is we will launch and be operating every year more AI in space than the cumulative total on Earth... a few hundred gigawatts per year of AI in space and rising. — Elon Musk, February 2026