The National Oceanic and Atmospheric Administration (NOAA) will migrate its Weather and Climate Operational Supercomputing System, known as WCOSS, to commercial cloud infrastructure, according to a NOAA announcement. Google Cloud will serve as the primary high-performance computing provider for the program. The agency's forecast models, including the Global Forecast System and Global Ensemble Forecast System, will move to the cloud environment in phases through early 2027, with full completion targeted for December 2027.
NOAA said the shift is intended to reduce reliance on fixed on-premises hardware, allow computing capacity to scale during periods of high demand such as tropical storm season, and speed the transition of research models into operational use. The partnership also extends to NOAA's AIGFS model suite, a new set of machine learning-based forecast models built on the Google DeepMind framework, which NOAA says can generate a global forecast in seconds rather than hours.
The migration follows contracts announced earlier in 2026 to bring cloud-based technology to forecasters at NOAA's 122 local Weather Forecast Offices and other National Weather Service centers. NOAA's weather models are developed and managed by the National Weather Service's Modeling and Development division in College Park, Maryland, part of the National Centers for Environmental Prediction. NOAA said it expects to be among the first operational numerical weather prediction agencies worldwide to run its forecasting systems on commercial cloud infrastructure.
The announcement builds on AI-driven approaches already reshaping operational climate and weather risk forecasting, and adds federal weather modeling to the list of workloads moving onto hyperscale cloud infrastructure that is already straining grid capacity in several U.S. markets. NOAA's reliance on a single commercial provider for its primary forecasting infrastructure also follows the broader pattern of public and private institutions consolidating critical computing workloads onto hyperscaler platforms.
