$NVIDIA(NVDA.US)-backed startup StarCloud successfully trained the first AI model on an "Orbital Data Center" — Jensen Huang's quote "Everywhere computing is done" now extends even to space.

1. What is an "Orbital Data Center"?

StarCloud launched satellite modules equipped with Nvidia's radiation-hardened aerospace-grade GPUs, completing AI model training (not just inference) onboard. Previously, training was thought to require massive power and cooling, making orbital operations nearly unthinkable — this breakthrough shatters that stereotype.

2. Why move to space?

Compelling economic and physical logic addresses two 2025 AI bottlenecks:

a. Cooling: Ground data centers consume vast water/electricity to cool chips. Space’s ambient −270°C provides abundant "free" cooling (though thermal engineering remains challenging).

b. Energy: Unobstructed solar in low-Earth orbit enables near 24/7 power with higher intensity, offering greener computing.

Deeper insight: Much AI data (weather, surveillance, comms) originates from satellites. Instead of bandwidth-limited downlinks for raw data, space-based training transmits only processed models — extending edge computing to orbit.

Additionally, space radiation cripples standard chips; Nvidia’s aerospace GPUs demonstrate material science and packaging advances, widening its lead.

$T-Rex 2X Long NVIDIA Daily Target ETF(NVDX.US)$GraniteShares 2x Long NVDA Daily ETF(NVDL.US)