Photo credit: venturebeat.com
Nvidia has introduced the Omniverse Blueprint for Earth-2, a cutting-edge initiative aimed at enhancing weather analytics and forecasting accuracy.
The significance of this development cannot be overstated, especially as the economic consequences of climate-related weather phenomena have reached an astounding $2 trillion globally over the past decade. Through this blueprint, Nvidia provides organizations with the necessary tools to bolster risk management and improve readiness for disasters.
The Omniverse Blueprint for Earth-2 comprises reference workflows that integrate Nvidia’s GPU acceleration libraries, a physics-AI framework, essential development tools, and microservices. This comprehensive package is designed to support businesses in transitioning from initial prototyping stages to full-scale production of weather forecast models.
During his keynote address at the GTC 2025 event, Nvidia’s CEO, Jensen Huang, emphasized the groundbreaking nature of Earth-2, describing it as a pivotal technological accomplishment in weather prediction. He noted that its implications extend beyond meteorology, signaling the scale of resources required to create an Earth-like metaverse or simulation environment.
Huang remarked, “This is one of the biggest developments of the past year. We are enabling the acceleration of all major CAE software through advanced libraries made possible by CUDA.”
Among the accessible components of the Nvidia Earth-2 blueprint are Nvidia NIM microservices, including CorrDiff for downscaling weather data, and FourCastNet for analyzing global atmospheric dynamics. These tools are actively employed by weather technology firms, research institutions, and government bodies to derive valuable insights and reduce the risks associated with extreme weather events.
“The frequency of extreme weather events and natural disasters is rising, endangering lives and infrastructure,” Huang stated. “The Nvidia Omniverse Blueprint for Earth-2 will empower industries globally to prepare for and address the challenges posed by climate change and weather-related calamities.”
Ecosystem Support
At Nvidia’s GTC 2024 event, several prominent climate technology firms, such as AI leader G42, JBA Risk Management, and Spire Global, have begun harnessing the blueprint to create innovative AI-enhanced solutions.
By leveraging proprietary enterprise data within the burgeoning $20 billion climate tech sector, the Nvidia Earth-2 platform is enabling developers to issue warnings and provide updates on weather forecasts in a matter of seconds, vastly outpacing the traditional CPU-driven modeling methods that can take much longer.
G42 is merging various elements of the Omniverse Blueprint with its advanced AI forecasting models for Earth-2 to equip the United Arab Emirates’ National Center of Meteorology with state-of-the-art technologies for enhanced weather prediction and disaster management.
“G42 is pushing the boundaries of AI-powered forecasting to help governments and businesses bolster their resilience against extreme weather in a dynamically changing environment,” stated Andrew Jackson, CEO of Inception, a subsidiary of G42. He further explained how their tailored AI system effectively translates broad weather data into highly localized forecasts, significantly improving prediction speed and accuracy. This collaboration with Nvidia’s CorrDiff model allows decision-makers to leverage actionable intelligence to protect communities, infrastructure, and plan for a more sustainable future.
In addition, Spire Global is utilizing the AI capabilities from the blueprint as a framework to produce new AI-driven products that amalgamate its satellite data and extend medium to long-term forecasts up to 45 days. These models, powered by Nvidia GPUs and the Earth-2 blueprint, function 1,000 times faster than conventional physics-based models, thus enabling comprehensive ensemble forecasting that encompasses a wide array of potential weather scenarios.
Other entities exploring or adopting Earth-2 include Taiwan’s Central Weather Administration, The Weather Company, 3D mapping firm Ecopia AI, spatial analytics leader ESRI, green-energy supplier GCL Power, flood management experts JBA Risk Management, aerospace innovator OroraTech, and Tomorrow.io, known for its resilience platform harnessing unique space data and weather intelligence.
Groundbreaking Generative AI for Climate Tech
The Earth-2 platform is equipped with an extensive range of tools, microservices, and state-of-the-art AI weather models, enabling the visualization and simulation of the Earth’s climate.
CorrDiff, which forms part of the Omniverse Blueprint, is offered as an Nvidia NIM microservice. It boasts remarkable efficiency, operating 500 times faster and 10,000 times more energy-efficient than traditional CPU-driven numerical weather prediction methods.
This blueprint empowers independent software vendors to create and implement AI-enhanced solutions while utilizing observational data to heighten accuracy and speed.
Esri, a foremost leader in geospatial technology, is partnering with NVIDIA to integrate its ArcGIS platform with Earth-2 via the blueprint. Similarly, OroraTech is examining the potential of connecting its data platform with the Omniverse Blueprint for Earth-2.
Tomorrow.io has contributed its near-real-time satellite data to support the development of an NVIDIA digital twin of Earth, facilitating the training, inference, and reinforcement of next-generation AI models.
A pivotal element of the blueprint is Nvidia Omniverse, which serves as a developmental platform for constructing OpenUSD-based 3D workflows and applications. The Omniverse Blueprint for Earth-2 exemplifies how developers can exploit Omniverse software development kits and microservices to build RTX-powered visualization pipelines for geospatial and weather-related data.
Nvidia DGX Cloud-Powered Compute
The Omniverse Blueprint for Earth-2 takes advantage of the Nvidia DGX Cloud platform, showcasing comprehensive acceleration for AI-enhanced weather forecasting.
Utilizing Nvidia DGX GB200, Nvidia HGX B200, and Nvidia OVX supercomputers, the blueprint paves the way for simulating and visualizing global climate models with exceptional speed and scale.
Source
venturebeat.com