Orbital compute · spacerack.si

The world's first startup launching entire AI pods into orbit.

Container-scale AI factories, powered by the Sun, cooled by space.

The moment

We take that idea off the planet.

Today NVIDIA introduced DSX POD — a complete AI factory inside a modular container: seven liquid-cooled Vera Rubin racks, deployable in 14 days instead of six months.

Form factor40-ft container
RacksSeven, liquid-cooled
Deploy14 days

NVIDIA, DSX POD and Vera Rubin are trademarks of NVIDIA Corporation. SpaceRack is independent.

01 · Giant solar wings

Power straight from the source.

Huge deployable solar arrays extend from both long sides of every pod. Continuous power from the Sun — no grid interconnect, no substation queue, no land.

  • Two wings per pod, four panels each
  • Unfold in orbit, minutes after separation
  • No grid. No land. No neighbours.

02 · Half-power by design

Every pod runs at 50% power.

Less heat to reject, lower energy draw per rack, and far longer hardware life. Headroom is a feature, not waste.

50%

03 · Evaporative cooling chamber

Heat goes to deep space.

A dedicated evaporation chamber, engineered for vacuum, carries heat away from the racks and rejects it to deep space through the radiator fin array on the pod's roof.

  • Evaporation chamber coupled to every rack loop
  • Radiator fins face away from the Sun
  • No water towers, no chillers, no coolant drawn from Earth

04 · Scale

0

pods per year.

We are ready to launch one thousand pods a year.

How it works

Four steps, Earth to orbit.

01

Build on Earth

Racks, power and cooling integrated inside one sealed container.

02

Launch as one unit

The pod flies whole, wings folded flat against its sides.

03

Unfold solar wings

Both arrays deploy panel by panel and lock to the Sun.

04

Compute in orbit

The pod joins the fleet and starts taking workloads.

Contact

Compute beyond the planet.

Operators, launch partners and investors: we are talking to all three.