Orbital, a space datacenter startup, is pursuing a 10,000-satellite constellation for distributed AI inference despite CEO Euwyn Poon admitting current launch economics are unviable. The plan hinges on SpaceX reducing costs from ~$7,000/kg to ~$10/kg, while assuming launch providers will sell spare capacity to competitors. Orbital is developing compact satellites (four-refrigerator-sized) housing 100kW of compute, with heat dissipation via radiators.
Infrastructure
Orbital datacenter startup CEO admits launch economics don't fly, presses ahead regardless
Orbital is pursuing a 10,000-satellite constellation for distributed AI inference despite the CEO admitting current launch economics are unviable, betting on a 700x cost reduction from SpaceX.
Wednesday, April 15, 2026 12:00 PM UTC2 MIN READSOURCE: The RegisterBY sys://pipeline
Tags
infrastructure
/// RELATED
ProductsApr 24
Show HN: Browser Harness – Gives LLM freedom to complete any browser task
Browser Harness library enables LLMs to autonomously complete browser tasks through self-healing architecture that lets agents write missing functions mid-execution.
InfrastructureApr 14
AI datacenter startup Fluidstack in talks for $1B round at $18B valuation months after hitting $7.5B, says report
Fluidstack's leap to $18B valuation, propelled by Anthropic's landmark $50B datacenter commitment, signals a fundamental consolidation—AI leaders are building dedicated, controlled compute to bypass cloud hyperscalers.