Orbital Targets 2027 Launch to Move AI Inference into Low Earth Orbit

Why this matters

For data and analytics teams: Off-grid compute capabilities could enable processing of massive global data streams without terrestrial bandwidth or capacity bottlenecks.

For AI/ML practitioners: The deployment of specialized Nvidia Space-1 Vera Rubin GPUs in orbit introduces a new hardware tier specifically optimized for space-based inference.

For business leaders: Moving compute to orbit decouples infrastructure scaling from terrestrial energy price volatility and increasingly stringent local grid regulations.

Orbital has announced it will launch its first test mission, Orbital-1, in April 2027 aboard a SpaceX Falcon 9 rocket. The Los Angeles-based startup aims to deploy a constellation of satellites that function as high-performance data centers in Low Earth Orbit (LEO). This infrastructure is designed to solve the critical energy and cooling shortages currently stalling AI development on the ground by utilizing continuous solar power and radiating heat directly into the vacuum of space. The company is positioning itself at the infrastructure layer of the AI stack, specifically targeting the inference market. Unlike the training of large language models, which requires thousands of GPUs to be tightly coupled with near-zero latency, inference tasks can be distributed across a network of independent nodes. Orbital's satellites will house clusters of Nvidia's recently announced Space-1 Vera Rubin modules, which are engineered for high-density processing in the harsh environment of LEO.

Context

This initiative arrives as terrestrial data center operators face unprecedented resistance from energy providers and environmental regulators. Microsoft recently signed a 20-year deal to restart the Three Mile Island nuclear plant to secure dedicated power, a move that highlights the desperation for stable energy sources. Orbital is effectively building a new category of 'sovereignty-as-a-service' where compute resides outside the physical and regulatory boundaries of any single nation-state.

Signal or noise?

For now, this is a high-conviction signal for the infrastructure sector but remains a long-term watch for software teams. The 2027 launch date means commercial availability is years away. The primary noise to filter out is the promise of immediate cost savings; space-based compute currently carries massive launch and miniaturization costs. The real signal to watch over the next 12 months is Orbital's success in radiation-hardening standard GPU architectures, which remains the single largest technical hurdle for sustained space-based compute.