Starcloud Hits $1.1B Valuation as It Accelerates Strategy for Orbital Data Centers

New updates have been reported about Starcloud.

Starcloud has reached a $1.1 billion valuation following a new $200 million Series A led by Benchmark and EQT Ventures, making it one of the fastest Y Combinator alumni to achieve unicorn status and positioning it as an early leader in space-based compute. The company is betting that orbital data centers will relieve terrestrial bottlenecks in energy, permitting, and geopolitics, though the model still relies on unproven hardware resilience in space and large upfront capex.

The company launched its first satellite with an Nvidia H100 GPU in November 2025, claiming the first deployment of a terrestrial-grade GPU in orbit and the first in-orbit AI training run, which also executed a version of Gemini. Starcloud plans to launch Starcloud 2 later this year with multiple GPUs, including Nvidia Blackwell, an AWS server blade, and a bitcoin mining rig, plus what it says will be the largest deployable radiator flown on a private satellite to manage thermal loads.

Strategically, the core bet is Starcloud 3, a three-ton, 200-kilowatt data center spacecraft designed to launch on SpaceX’s Starship using the same “PEZ dispenser” deployment system as Starlink satellites. CEO Philip Johnston projects that, assuming launch prices fall to about $500 per kilogram, Starcloud 3 could deliver orbital compute at roughly $0.05 per kWh, approaching parity with terrestrial data centers and opening a path for full-scale distributed space cloud services.

Execution risk centers on launch economics and timing, as Starship is not yet commercially operational and Johnston does not expect broad availability until 2028–2029, delaying true cost competitiveness. Until then, Starcloud will continue deploying smaller platforms on Falcon 9 and focus on higher-value workloads, such as on-orbit processing for other spacecraft, illustrated by its current work analyzing radar imagery for Capella Space and other potential inference-heavy applications.

Johnston describes a dual business model: near term, selling compute cycles and data processing to satellites already in orbit; longer term, shifting part of data center workloads off Earth once launch costs and on-orbit infrastructure mature. Early experience has been mixed but instructive, with one Nvidia A6000 failing during launch, giving Starcloud critical engineering data on survivability, power management, cooling, and synchronization for future generations.

The company is operating in an emerging competitive field that includes Aetherflux, Google’s Project Suncatcher, and Aethero, and it must also factor in SpaceX’s proposal to deploy up to one million compute satellites for its own AI and automotive workloads. Johnston argues Starcloud can coexist with SpaceX by positioning as an energy and infrastructure provider and by targeting broader third-party cloud use cases rather than tightly integrated, vertically owned workloads, giving investors and partners a differentiated thesis on orbital compute.