Fight the environment
- High water demand or complex closed-loop cooling
- Exposure to storms, floods, fires, grid congestion, and politics
- Heat, noise, and land-use conflict near communities
- Expansion tied to local permitting and power availability
Compute beyond weather, borders, and heat islands
The future of data centers belongs on the Moon.
Darkside Data Center is a proposal for high-density lunar compute campuses: shielded under regolith, powered by continuous ridge-line solar, and cooled by the cold vacuum that terrestrial infrastructure spends billions trying to simulate.
-173 C
Night-side thermal sink
0 weather
No storms, floods, or wildfires
14 days
Lunar daylight per cycle
~2.6 s
Earth-Moon light round trip
Why leave Earth
Terrestrial campuses are running into physics, permitting, and public trust.
AI-scale compute wants cheap power, extreme cooling, geopolitical stability, and room to expand. On Earth, those needs collide with water rights, transmission constraints, heat rejection, land use, noise, and the fragility of regional grids.
A lunar facility changes the constraint set. The far side offers isolation from radio interference, no weather-driven downtime, no biosphere to heat or drain, and a vacuum environment that can make radiative cooling the primary thermal design instead of an expensive afterthought.
Lunar vs terrestrial
Same compute mission. A radically different operating envelope.
Design with the environment
- Vacuum-compatible heat rejection through large radiator fields
- Regolith shielding for radiation, micrometeoroids, and thermal mass
- Solar-rich siting near high-illumination ridges and polar resources
- Remote, modular growth without competing for local water or land
How it works
A lunar data center is not a warehouse in space. It is an infrastructure stack.
Shielded compute vaults
Server modules sit in pressure-stabilized vaults covered by meters of sintered regolith for radiation and impact protection.
Radiator farms
Heat pipes move waste heat to articulated radiator fields that face deep space, turning vacuum into a thermal advantage.
Ridge-line power
Solar arrays, fuel cells, and stored hydrogen bridge lunar night while high-illumination sites keep generation predictable.
Optical relay mesh
Laser links route traffic through lunar orbit and ground stations, reserving local inference and batch jobs for latency-tolerant workloads.
Operating model
Start with the workloads that already tolerate distance.
Lunar compute does not need to replace every data center. It starts where latency matters less than capacity, resilience, and energy certainty: model training, scientific simulation, archival processing, rendering, cryptographic workloads, and autonomous lunar operations.
Training runs can checkpoint locally, sync gradients in scheduled windows, and use lunar power/cooling advantages for sustained high-density operation.
Deployment path
Build it like critical infrastructure, one certifiable layer at a time.
-
Phase 1
Orbital proof
Validate optical networking, hardened servers, autonomous repair routines, and radiation-tolerant storage in lunar orbit.
-
Phase 2
Surface pilot
Land containerized compute, regolith shielding equipment, compact radiators, and a robotic maintenance bay.
-
Phase 3
Polar campus
Scale power fields, thermal loops, spare-part manufacturing, and Earth-lunar traffic scheduling for commercial workloads.
Mission brief