**Trader sentiment on the "AI data center in space by...?" market reflects accelerating momentum from AI-driven terrestrial constraints alongside persistent technical and economic hurdles for full-scale orbital deployments.** Explosive demand for large language model training and inference has strained Earth-based power grids and land availability, prompting major players to explore space for abundant solar energy and passive cooling. NVIDIA’s March 2026 launch of radiation-hardened Space-1 Vera Rubin modules and related platforms directly targets orbital AI workloads, while its earlier support for Starcloud enabled the first in-orbit large language model training on an H100 GPU in December 2025. Google’s Project Suncatcher, unveiled in November 2025, targets prototype satellite tests by early 2027 with high-speed laser links, and SpaceX’s FCC filing for up to one million AI compute satellites—with potential first launches around 2028—adds credible launch capacity. Smaller milestones, such as Axiom Space nodes on the ISS and Orbital’s planned 2027 inference-focused test, demonstrate progress on edge AI but fall short of the market’s typical threshold of at least 100 data-center-grade accelerators in a single orbital platform. Competitive dynamics favor integrated efforts (SpaceX/xAI, Google/Planet Labs) over pure startups, yet radiation hardening, thermal management, and launch economics remain key swing factors. Upcoming catalysts include 2027 prototype flights and any confirmed partnerships that could shift implied probabilities for operational clusters before 2028.