New AI Model Captures the Milky Way in Stunning Detail

Explore

HAdorned with more than 100 billion stars, including our Sun, the vast Milky Way is naturally difficult to simulate with computer models. The most sophisticated models available can only simulate the mass of about 1 billion suns, meaning that the smallest resolution unit of these images still represents an average of about 100 stars, resulting in a relatively fuzzy picture. Now, in a major step in AI-assisted computer modeling, researchers have created a simulation of 100 billion stars in the Milky Way, capable of tracking individual stars.

This breakthrough was achieved with the help of a deep learning AI surrogate that helped researchers overcome a particularly thorny sticking point: the behavior of supernovas. Simulating the fine particles released from these exploding stars into space creates a heavy computational burden and headaches for human researchers. By feeding surrogate high-resolution supernova simulations to deep learning, a team led by Keiya Hirashima of the RIKEN Center for Interdisciplinary Theoretical and Mathematical Sciences was able to teach the program to predict how gas from a supernova would propagate 100,000 years into the future. With the AI ​​taking care of the little things, the rest of the simulation was freed up to focus on the bigger picture.

The result is a new model that not only includes 100 times more stars than the best simulations to date, but was also generated more than 100 times faster. The model was recently presented at SC ’25, an international conference on supercomputing.

Read more: »The galaxy has become too big»

Looking ahead, Hirashima and his team have high hopes for their new hybrid modeling technique. “I believe that the integration of AI with high-performance computing marks a fundamental shift in how we approach multi-scale, multi-physics problems in computer science,” Hirashima said in a statement. “This achievement also shows that AI-accelerated simulations can go beyond pattern recognition to become a true tool for scientific discovery, helping us trace how the elements that formed life itself emerged within our galaxy.”

In addition to modeling galaxies, their new approach also promises to elucidate more terrestrial phenomena, including oceanography, meteorology and climate change.

Enjoy Nautilus? Subscribe for free to our newsletter.

Main image: NASA/JPL-Caltech/ESA/CXC/STScI

• Jake Currie

  Published on November 18, 2025

  Jake Currie is a writer based in Brooklyn, New York.