With the launch of Aurora, we began providing researchers with access to one of the world’s most powerful supercomputers for open science. After years of planning and preparation, it has been exciting to see how our users are taking advantage of the system’s advanced capabilities for large-scale simulation, AI, and data analysis.
Scientists across disciplines are using Aurora to accelerate research in areas ranging from biology and materials science to cosmology and energy technologies, tackling complex problems with greater speed, scale, and fidelity. Many of these early efforts demonstrate how simulation, AI, and data science methods can be combined to drive breakthroughs in science and engineering.
This year also marked an important step in how we facilitate and support science. As experiments and simulations generate increasingly large volumes of data, the traditional model of accessing supercomputers through batch jobs is no longer sufficient for many workflows. Building on decades of experience in this space, we are delivering a Service-Enabled Science program that makes these capabilities broadly accessible to a new generation of users. This approach brings together HPC, AI, data, and workflow tools as integrated services that support the full research lifecycle.
This approach is already transforming how science is conducted. Through our close integration with the Advanced Photon Source and other experimental facilities, researchers can now analyze data in near real time, refine experiments as they unfold, and move more quickly from observation to insight. By making computing an integrated part of the scientific process rather than a separate step, we are helping streamline discovery.
A key milestone in this effort was the launch of the ALCF Inference Service, which provides secure, scalable access to a range of large language models for scientific applications. By enabling researchers to run large-scale inference workloads directly on ALCF systems, the service supports everything from experiment-time analysis to high-throughput data processing and AI-driven workflows. Just as importantly, it gives the research community a stable, transparent environment for applying AI at scale.
These efforts align closely with the U.S. Department of Energy’s Genesis Mission, a new national initiative to accelerate scientific discovery through AI. At Argonne, ALCF contributes to this vision through its expertise and computing resources, including Aurora and the Inference Service, as well as next-generation systems such as Tara, Minerva, and Janus that support the broader user community. In parallel, we are engaged in public-private partnerships to develop new systems, including Equinox and Solstice, to further advance the goals of the Genesis Mission. Together, these efforts are helping to lay the foundation for a more connected, AI-enabled scientific ecosystem.
Workforce development remains central to our mission. In 2025, we expanded our efforts to help researchers and students adopt these new approaches, including the launch of our Service-Enabled Science training series. Through instructional webinars and real-world examples, we are working to build a community that can fully utilize our integrated HPC and AI services to enhance research. To help build an AI-ready workforce for the future, we also continued our AI for Science training series, focusing on advanced topics such as large-scale model training, inference workflows, and coupling simulations with AI.
The achievements highlighted in this report reflect the dedication and expertise of the ALCF team, as well as the strong partnerships we maintain across DOE, industry, and the broader research community. We are proud to support this work and to share these advances with you.
Looking ahead, we are excited to build on this momentum by deploying new systems, expanding our services, and continuing to meet the evolving needs of science in the era of exascale computing and AI.