When does orbit beat these rates?

Orbital GPU capacity reaches the market in 2027. The question isn't whether orbit beats your cloud bill today — it's when it will for your workload. This calculator gives you that number, as a modeled estimate. No commercial orbital capacity is on the market yet, so orbital cost here is a modeled figure (~3× the cheapest terrestrial rate, per public space-datacenter cost models): your workload's terrestrial cost across AWS, GCP, and Azure at modeled public rates, the modeled orbital premium, and the exact $/GPU-hr at which orbit reaches parity.

Where these numbers come from

Terrestrial rates: modeled public on-demand rate card (approx. March 2026) — H100: GCP $3.98 / AWS $4.15 / Azure $4.30 per GPU-hr; A100: GCP $2.10 / AWS $2.21 / Azure $2.35; L40S: GCP $1.15 / AWS $1.22 / Azure $1.30. Orbital: modeled at 3.0× the cheapest terrestrial rate — a modeled assumption ahead of the first commercial capacity expected in 2027, not a contracted rate. Every figure on this page is modeled, not a quote.

Full derivation — the rate card's basis, the 3.0× multiple, the physics behind every routing figure, and the public endpoint that reproduces each one: /methodology.