4.1 KiB
| type | title | author | url | date | domain | secondary_domains | format | status | priority | tags | extraction_model | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| source | Starcloud Raises $170M Series A at $1.1B Valuation — Roadmap to Starcloud-2 and Starcloud-3 | TechCrunch (@TechCrunch) | https://techcrunch.com/2026/03/30/starcloud-raises-170-million-series-ato-build-data-centers-in-space/ | 2026-03-30 | space-development | article | null-result | high |
|
anthropic/claude-sonnet-4.5 |
Content
Starcloud announced a $170M Series A at a $1.1B valuation on March 30, 2026, led by Benchmark and EQT Ventures. Total raised: $200M+. Fastest YC graduate to reach unicorn status.
Starcloud-2 (October 2026 launch target):
- Multiple GPUs including NVIDIA Blackwell chip
- AWS server blade
- Bitcoin mining computer (!)
- "Largest commercial deployable radiator ever sent to space"
- 100x the power generation of Starcloud-1
- First satellite to run commercial edge/cloud workloads for paying customers
- Early customers: Crusoe (AI compute startup)
- Partners: AWS, Google Cloud, NVIDIA
Starcloud-3 (development phase, post-Starcloud-2):
- 200 kW capacity
- 3 tonnes spacecraft
- Fits SpaceX's "PEZ dispenser" Starship deployment system
- CEO Philip Johnston: "first orbital data center that is cost-competitive with terrestrial data centers"
- Target: $0.05/kWh
- CONDITION: requires commercial launch costs ~$500/kg
CEO direct quote on cost threshold: expects Starcloud-3 to be competitive IF launch costs reach ~$500/kg. Notes that "commercial Starship access isn't expected until 2028-2029" — meaning cost-competitive ODC at scale is a 2028-2030 story at earliest.
Number of advanced GPUs currently in orbit as of 2026: "numbered in the dozens" (vs. ~4 million H100s sold to terrestrial hyperscalers in 2025).
Agent Notes
Why this matters: This is the most specific and authoritative data point connecting ODC cost competitiveness to a specific launch cost threshold. CEO explicitly says: competitive at $500/kg. Current Starship commercial pricing: ~$600/kg (Voyager Technologies filing). The gap is real but narrow — this could clear in 2027-2028 with higher reuse cadence.
What surprised me: The Starcloud-2 manifest includes a bitcoin miner. This is a signal that ODC economics are not just AI — any computation that benefits from free solar power, zero cooling costs (well, radiator costs), and proximity to orbital infrastructure is a candidate. Bitcoin mining in space is wild but consistent with the power-cost-arbitrage logic.
What I expected but didn't find: Specific performance numbers for Starcloud-2's compute capability (FLOPS, watts of compute vs. watts total). The "100x power generation" metric suggests Starcloud-2 is maybe 1-2 kW of compute power (Starcloud-1 is likely <100W of compute). This is still toy scale vs. terrestrial data centers.
KB connections: This source contains the clearest real-world evidence for the launch cost keystone claim. $500/kg = ODC industry activates. $600/kg = ODC industry doesn't. This is Belief 2 operating exactly as the threshold model predicts.
Extraction hints:
- CLAIM CANDIDATE (HIGH VALUE): Starcloud-3's cost competitiveness threshold of $500/kg launch cost is the first explicitly stated industry activation threshold for orbital data centers — directly instantiating the general claim that each launch cost milestone activates a new industry.
- Note the 3-year satellite lifecycle in Starcloud-1 (11 months at 325km). The cost model assumes longer lifetimes at higher orbits — but radiation environment is harder there.
Curator Notes
PRIMARY CONNECTION: launch cost reduction is the keystone variable that unlocks every downstream space industry at specific price thresholds — this source is the most explicit evidence for that claim in a specific industry context with a specific dollar figure. WHY ARCHIVED: Contains the key empirical validation of the launch cost threshold model for the ODC industry. The $500/kg threshold is citable and specific. EXTRACTION HINT: Extract the threshold claim first, then the radiator-as-binding-constraint observation second.