Photo: Spire

Spire Global introduced two AI weather models: AI-WX and AI-S2S, developed on the Nvidia Omniverse Blueprint for Earth-2 platform, Spire announced on March 18.

These models utilize Spire’s satellite data to generate medium-range and sub-seasonal forecasts, providing prediction capabilities for energy, utilities, and commodities sectors.

Spire said the AI-WX model offers forecasts up to 20 days in advance, while AI-S2S extends predictions to 45 days. The models integrate data from Spire’s satellite network, including atmospheric profiles, soil moisture readings, and ocean surface winds to produce probabilistic forecasts, presenting a spectrum of potential weather scenarios. 

Michael Eilts, general manager of Weather and Climate at Spire, commented: “By harnessing the computational power of NVIDIA graphic processing units (GPUs) paired with the unique space-based data from our satellite network, we have developed AI-driven models that transform how industries manage weather risks.”

Spire said the purpose of its AI models is to address challenges by delivering insights that allow companies to anticipate and mitigate weather-related disruptions. Freepoint Commodities and Dominion Energy are some of the companies that use Spire tools to assess weather risks.

The launch of these models adds to Spire’s previous collaboration with Nvidia in 2024, which integrated Spire’s radio occultation data with Nvidia Earth-2 Cloud APIs.