IBM & NASA Launch AI Digital Twin of the Sun: "Surya"

IBM and NASA have unveiled Surya, the first-of-its-kind AI foundation model that creates a digital twin of the Sun, revolutionizing our ability to predict dangerous solar flares and space weather events. Released on August 20, 2025, this groundbreaking model can forecast solar activity up to two hours in advance with unprecedented accuracy.

"With Surya we have created the first foundation model to look the Sun in the eye and forecast its moods."

— Juan Bernabé-Moreno, IBM Director of NASA Scientific Collaboration

Named after the Sanskrit word for "Sun," Surya represents a quantum leap in space weather forecasting. The model addresses a critical gap in our ability to predict solar storms that can disrupt satellites, power grids, GPS systems, and communications infrastructure across Earth.

The Solar Threat Challenge

Solar flares have the potential to disrupt power grids, GPS, and even the internet. During the Sun's most active phases, this scenario has the potential to play out hundreds of times each month. Traditional prediction methods provide only one hour of advance warning, making Surya's two-hour forecasting capability a significant breakthrough.

Surya is a 366-million-parameter foundation model that processes raw solar imagery from NASA's Solar Dynamics Observatory (SDO). The model uses a long-short vision transformer with a spectral gating mechanism to handle the observatory's high-resolution 4096 x 4096-pixel images.

Unlike traditional AI systems that require extensive labeling, Surya can adapt quickly to new tasks and applications. The model learns solar physics patterns directly from raw observational data, making it highly versatile for various space weather prediction tasks.

Surya's breakthrough capabilities stem from NASA's Solar Dynamics Observatory, which has been continuously monitoring the Sun since 2010. The observatory has provided an unbroken, high-resolution record of the Sun for nearly 15 years through capturing images every 12 seconds in multiple wavelengths, plus precise magnetic field measurements.

Training Methodology

Researchers took a nine-year slice of SDO data, first harmonizing the different data types, then experimenting with AI architectures to process it. Rather than traditional image reconstruction, they trained Surya to predict future solar states, enabling genuine forecasting capabilities.

"This is an excellent way to realize the potential of this data — pulling features and events out of petabytes of data is a laborious process and now we can automate it."

— Kathy Reeves, Solar Physicist, Harvard–Smithsonian Center for Astrophysics

Surya addresses multiple critical space weather prediction tasks that affect modern technological infrastructure. The model's applications span from immediate operational warnings to fundamental solar physics research.

Space Mission Applications

Astronauts bound for the Moon or Mars may need to depend on precise predictions to shelter from intense radiation during solar particle events. Surya's enhanced prediction capabilities are crucial for protecting human space exploration missions.

IBM and NASA have made Surya freely available to the global research community, democratizing access to advanced solar prediction capabilities. The model and associated tools are available through multiple platforms for maximum accessibility.

SuryaBench Dataset

Along with the model, researchers have released SuryaBench, a comprehensive set of curated datasets and benchmarks designed to accelerate space weather research and solar physics studies. Each application is like a bridge to help the scientific community cross to the other side of the river, enabling researchers worldwide to build upon this foundation.

Surya represents just the beginning of AI-driven solar physics research. The model's foundation architecture enables adaptation to other celestial bodies and space weather phenomena, potentially revolutionizing our understanding of stellar behavior across the universe.

Broader Scientific Applications

"Understanding the sun is a proxy for understanding many other stars," Bernabe-Moreno says. The methodologies developed for Surya could be applied to study other stellar systems and advance our understanding of space weather throughout the galaxy.

"We are advancing data-driven science by embedding NASA's deep scientific expertise into cutting-edge AI models. This model empowers broader understanding of how solar activity impacts critical systems and technologies that we all rely on here on Earth."

— Kevin Murphy, Chief Science Data Officer, NASA Headquarters