Space Solar Array Specific Power

Answer

Optimistic: 300 W/kg -- Assumes next-generation thin-film or ultra-lightweight flexible arrays (e.g., Starpath Starlight Air-class technology with deployment structures) reach production readiness by 2030-2032, with high-efficiency cells on minimal-mass substrates. Not yet flight-proven at system level.

Central: 150 W/kg -- Represents the high end of current state-of-the-art rigid/semi-rigid deployable arrays and the demonstrated performance tier of advanced flexible arrays like FOSA (140 W/kg flight-proven) and optimized ROSA derivatives. This is the value most orbital compute analyses assume for near-term (2028-2032) deployments.

Conservative: 80 W/kg -- Uses flight-proven flexible array performance (ROSA at 100 W/kg cell-level, derated for deployment mechanism overhead, wiring harness, and design margins). Reflects what can be procured and flown today at constellation scale with minimal technical risk.

Evidence

Flight-Proven Specific Power (Heritage Systems)

E1 [evidence:nasa-smallsat-power-soa.1] NASA's dataset of flown missions clusters around ~30 W/kg specific power, with empirical bounds of 1 W/kg (minimum) to 200 W/kg (maximum observed). This 30 W/kg figure includes all types of missions (CubeSats through large GEO platforms) and represents the "typical" rather than the "best achievable."

E2 [evidence:nasa-smallsat-power-soa.2] ROSA (Roll-Out Solar Array) achieves 100 W/kg as a flexible PV blanket system at 1000W capacity. ROSA has extensive flight heritage: ISS demonstration (2017), six IROSA wings deployed on ISS, DART mission, and Ovzon-3 GEO satellite (Jan 2024). It is the current gold standard for large-scale deployable power.

E3 [evidence:nasa-smallsat-power-soa.3] ExoTerra's FOSA (Flexible Origami Solar Array) achieves 140 W/kg at 150W peak BOL power. FOSA is listed in the NASA SOA database with flight heritage.

E4 [evidence:nasa-smallsat-power-soa.4] MMA Design's Hawk/zHawk deployed rigid (PCB-based) arrays achieve 95-121 W/kg in the 36-310W range.

E5 [evidence:nasa-smallsat-power-soa.5] DHV Technologies CubeSat panels achieve 42-140 W/kg depending on substrate (polyimide vs CFRP).

E6 [evidence:herasimenka-starlink-solar.1] Reverse-engineering analysis of Starlink Gen 1.x solar arrays estimates 78-100 W/kg specific power. The array uses 6-busbar 166mm silicon half-cells at ~18% efficiency, generating ~7,535 W total from 36 modules. Array mass estimated at 40-60 kg (module only) or 70-90 kg (with deployer). The author assesses this as "comparable to the best existing flex arrays using III-V multijunction cells."

State-of-the-Art and Near-Term Systems

E7 [evidence:nasa-smallsat-power-soa.6] State-of-the-art rigid arrays achieve up to 200 W/kg. This represents the empirical upper bound from the NASA dataset but is not the typical mission value.

E8 [evidence:dwarkesh-space-gpus.1] Dwarkesh Patel's mass budget analysis notes: "There are apparently companies that are targeting next gen thin film that reaches upwards of 500 W/kg, but the state of the art is 150 W/kg, and most missions right now fly 30 W/kg." He assumes 200 W/kg for his orbital datacenter analysis, characterizing this as "generous."

E9 [evidence:satnews-fractal-lab-iii.1] The Fractal Lab Part III analysis presents a three-tier technology maturity framework: flown heritage at ~30 W/kg, laboratory demonstrated at up to 200 W/kg, and near-term projection of ~100 W/kg for 2030s deployable systems at megawatt scale. It concludes that 100 W/kg "is a reasonable projection, aggressive relative to heritage systems but conservative relative to laboratory demonstrations."

E10 [evidence:mccalip-space-dc.1] McCalip's orbital data center economic model assumes 36.5 W/kg for Starlink-class (V2 Mini) solar arrays. He provides a sensitivity range from ISS heritage (3 W/kg) to advanced concepts (100-500 W/kg). This 36.5 W/kg figure is conservative and based on publicly available Starlink satellite mass/power ratios.

E11 [evidence:peraspera-realities.1] Per Aspera's orbital compute analysis states: "Highly optimized arrays can achieve ~150 W/kg (~6.7 kg/kW)." Multi-junction gallium arsenide cells at ~30% efficiency produce ~200 W/m^2 at AM0 irradiance (after accounting for incidence angle, temperature effects, etc.).

Next-Generation and Announced Technologies

E12 [evidence:starpath-solar-panels.1] Starpath Space's Starlight Air panels: 73 g/m^2 areal mass, ~$15/watt. Uses photovoltaic crystalline structure measured in "hundreds of nanometers" printed onto substrate fabric. 50 MW production facility planned; first deliveries 2026.

E13 [evidence:terawatt-starlight-specs.1] Starlight Air specifications: 16% efficiency, 73 g/m^2, fully flexible, rad-hard rated for LEO through Mars. At AM0 irradiance (1,361 W/m^2), this yields ~218 W/m^2 and a cell-level specific power of ~2,980 W/kg. However, this is bare-cell mass only; any deployment structure, wiring harness, and power conditioning would add substantial mass.

E14 [evidence:terawatt-starlight-specs.2] Starlight Classic specifications: 19% efficiency, 900 g/m^2, semi-flexible, silicon cells, $11.20/W. At AM0, this yields ~259 W/m^2 and cell-level specific power of ~287 W/kg.

E15 [evidence:nasa-smallsat-power-soa.7] Multi-junction III-V cells (AZUR 4G32-Adv, Rocket Lab IMM-alpha, SpectroLab XTE-SF/HF) achieve 31.5-32.2% BOL efficiency at AM0. Fraunhofer Institute reported up to 38% efficiency for four-junction designs under laboratory conditions.

E16 [opinion:handmer-2025-tweet.1] Casey Handmer proposes that in the limit, the specific power of orbital compute satellites approaches "the thinnest solar arrays that can be flown, probably close to 1 kg/m^2." At 1 kg/m^2 with reasonable cell efficiency, one Starship launch could deliver ~30 MW of orbital power generation.

E17 [evidence:mccalip-space-dc.2] McCalip's model uses 2.5% annual solar panel degradation as the nominal assumption, with sensitivity range of 1% (shielded designs) to 6% (unshielded) to 12% (polar/high-radiation orbits).

Degradation in LEO

E18 [evidence:mdpi-leo-degradation.1] Degradation modeling for LEO shows silicon solar cell power output decreases approximately 12.5% at 300 km and 7.8% at 700 km over six months. The dominant degradation mechanisms at 300-700 km altitude include trapped charged particles, atomic oxygen, and UV radiation. Higher altitudes within LEO experience less atomic oxygen erosion but more radiation exposure.

E19 [evidence:nasa-smallsat-power-soa.8] CIGS (copper indium gallium selenide) thin-film cells have demonstrated power conversion efficiencies up to 22.7%, offering a potential mid-range option between silicon (18-20%) and multi-junction III-V (31-32%) at lower cost.

AM0 Irradiance and Efficiency Context

E20 [evidence] Solar irradiance at AM0 (above atmosphere) is 1,361 W/m^2, approximately 36% higher than the 1,000 W/m^2 standard test condition (STC) used for terrestrial panels. This means space solar panels of equal efficiency produce ~36% more power per unit area than their terrestrial rating would suggest.

E21 [evidence:hn-xai-spacex-solar.1] HN discussion consensus: atmospheric derating brings insolation from ~1,367 W/m^2 (AM0) to ~1,000 W/m^2 at ground level (STC), representing a ~27% loss. Combined with no weather, no night (in SSO), and no seasonal variation, space solar panels achieve roughly 5-8x the average power output of identical ground panels, though the peak instantaneous difference is only ~36%.

Analysis

Technology Tiers

The evidence reveals a clear stratification of solar array specific power:

Deployable vs Rigid Trade-offs for Large-Scale Orbital Compute

For orbital compute at the 100+ kW per satellite scale, deployable flexible arrays are strongly preferred because:

1. Stowed volume -- Starship payload volume is finite; flexible arrays achieve 40x greater kW/m^3 stowed volume than rigid panels [nasa-smallsat-power-soa.2], enabling far more power capacity per launch.
2. Specific power -- Flexible arrays (ROSA 100, FOSA 140 W/kg) substantially outperform typical rigid arrays (30-60 W/kg at system level), reducing the mass fraction consumed by power generation.
3. Scalability -- ROSA's modular design (demonstrated at 20+ kW on ISS) scales naturally for the 50-200 kW range relevant to compute satellites.

The trade-off is that flexible arrays have lower cell efficiency (often using silicon at 18-20% rather than III-V at 30%+) and may be more susceptible to handling damage during manufacturing and deployment. For orbital compute, where launch mass is the binding constraint, the W/kg advantage of flexible arrays outweighs the W/m^2 advantage of high-efficiency rigid arrays.

Starlink as Benchmark

Starlink Gen 1.x achieves 78-100 W/kg using mass-produced silicon cells [herasimenka-starlink-solar.1], demonstrating that constellation-scale manufacturing can approach the performance of specialized aerospace solar arrays at much lower cost. Starlink V2 Mini at ~36.5 W/kg whole-satellite specific power [mccalip-space-dc.1] reflects a different optimization (the satellite does many things beyond power generation), but the solar module itself performs at the 78-100 W/kg tier.

Starlink v3 satellites are larger (~2,000 kg, ~250 m^2 solar area) but specific power data for the v3 solar array is not publicly available. If the v3 array achieves even modest improvement over Gen 1.x through thinner substrates or larger module format, 100-120 W/kg is plausible.

Degradation Considerations

Solar array degradation in LEO at 500-700 km is a meaningful factor over the 5-year operational lifetime assumed for orbital compute satellites:

• With coverglass protection: ~1-3% per year for III-V cells, ~2-4% per year for silicon cells [mccalip-space-dc.2]
• Without adequate protection: 6-12%+ per year [mdpi-leo-degradation.1]
• Sun-synchronous orbit advantage: Dawn-dusk SSO reduces time in Earth's radiation belts, lowering particle-induced degradation relative to other LEO inclinations.

Degradation affects effective specific power over the satellite lifetime. A 150 W/kg BOL array degrading at 2.5%/year delivers an average of ~141 W/kg over 5 years (geometric mean). This is factored into the scenario values.

Manufacturing Readiness

• ROSA (100 W/kg): TRL 9. Flight-proven at scale on ISS and commercial satellites. Production by Redwire Space.
• FOSA (140 W/kg): TRL 7-8. Limited flight heritage. Production by ExoTerra.
• Starlink modules (78-100 W/kg): TRL 9 at massive scale. SpaceX internal production.
• MMA FlexArray: TRL 6. Ground-qualified. Uses silicon cells on thin-film membranes.
• Starpath Starlight Classic (~287 W/kg cell-level): Pre-production. First deliveries planned 2026. No flight heritage.
• Starpath Starlight Air (~2,980 W/kg cell-level): Pre-production. 50 MW facility planned. No flight heritage.
• Perovskite thin-film: Lab demonstrations at extraordinary specific power (50 W/g = 50,000 W/kg cell-level). Not space-qualified. Stability and durability unproven in space environment.

Scenario Rationale

Conservative (80 W/kg): Uses ROSA-class technology (100 W/kg cell-level) with a 20% derating for deployment mechanisms, wiring harness, PCDU integration, and design margins. This represents technology that can be procured today from multiple vendors with flight heritage. It is pessimistic in the sense that Starlink already achieves 78-100 W/kg at massive scale, but accounts for the additional mass of thermal management integration, power conditioning, and structural margins needed for a compute-optimized (rather than telecom-optimized) satellite.

Central (150 W/kg): Assumes FOSA-class (140 W/kg) or next-generation ROSA derivative at the array level, with modest improvement from ongoing cell and substrate optimization by 2028-2030. This aligns with the "state-of-the-art" characterization in multiple sources [peraspera-realities.1, dwarkesh-space-gpus.1] and represents the planning assumption used by most serious orbital compute analyses. Achievable with III-V cells on lightweight flexible substrates, or with high-efficiency silicon on ultra-thin blankets.

Optimistic (300 W/kg): Assumes Starpath Starlight Classic-class technology (287 W/kg cell-level) reaches production maturity and that deployment mechanism mass is kept to a modest fraction through innovation in ultra-lightweight booms or tensioned membrane deployment. Alternatively, achieved through next-generation III-V cells (32%+ efficiency) on sub-100 g/m^2 substrates. This requires technology that is announced but not yet flight-proven, with manufacturing readiness expected no earlier than 2028-2030. The 500+ W/kg targets cited by some sources are excluded from even the optimistic scenario because they lack any near-term path to system-level demonstration.