When scientists at the US Department of Energy’s Argonne National Laboratory powered up the Aurora supercomputer, they were not just activating an advanced piece of hardware. They were launching a new era of AI-enabled scientific discovery.
Developed in partnership with Intel and Hewlett Packard Enterprise, Aurora is one of the most powerful computing systems ever built. With exascale performance, capable of carrying out more than two billion billion calculations per second, it sits at the convergence of artificial intelligence and high-performance computing. Its design and capabilities mark a pivotal shift in how the scientific community models reality, tests hypotheses and unearths insights.
Transforming science through machine intelligence
Aurora’s true significance lies not in raw processing speed but in its ability to blend simulation and machine learning at scale. From advanced climate modelling and aerospace engineering to understanding cancer biology and mapping neural networks, AI-infused simulations are helping scientists work faster, more accurately, and with greater clarity.
This hybrid approach is exemplified by the Trillion Parameter Consortium, where Aurora is being used to develop some of the largest open foundation models for science, including AuroraGPT. These models, trained across massive datasets, aim to capture complex interactions in biological, chemical, and physical systems that would otherwise take years—or decades, to understand through conventional methods.
As Olivier Franza, Aurora’s principal investigator at Intel, puts it, “Aurora is a phenomenal research engine that allows us to model complex physics and push scientific frontiers like never before.” That includes not only accelerating discovery but also increasing transparency and reproducibility by allowing researchers to test AI-generated predictions against known physical laws and outcomes.
A test of endurance and design
Aurora’s construction was far from straightforward. It required novel architectural decisions, cutting-edge accelerators, high-bandwidth memory, and software stacks that could scale without compromise. Covid-era supply chain issues, thermal design challenges, and the unpredictability of debugging at this level all tested the limits of the team involved.
At the centre of the effort was a deep collaboration between Intel’s engineers and Argonne’s researchers. Working side by side in lab basements and across time zones, they embedded a culture of responsiveness and trust that would ultimately prove critical. The project also marked a turning point for Intel, which moved beyond its traditional hardware manufacturing roots to become a systems integrator capable of delivering mission-critical infrastructure.
Bill Wing, Intel’s lead programme manager for Aurora, reflects that the project demanded “resilience at every level”, while also reshaping what leadership looks like under pressure. “True leadership means standing shoulder-to-shoulder, listening, and lifting others up,” he says.
A platform for the next generation of discovery
While the ribbon has only just been cut, Aurora is already reshaping how problems are approached across fields as diverse as quantum computing, energy systems, and national security. The success of the system, and the AI applications it supports, demonstrates how public-private collaboration can create platforms that serve both scientific ambition and public interest.
At a time when the limits of scale, transparency, and trust in AI are under scrutiny, Aurora offers a working model of what responsible innovation can look like. Its open-source foundations, transparent governance, and scientific purpose present a sharp contrast to the commercial race for closed, centralised AI power.
The project’s legacy may be as much about the process as the product. As Franza notes, “No one can take away the experience of building something of this scale, even if the road was rough.” Aurora is not just a machine; it is a commitment to science, to collaboration, and to the belief that when AI and HPC are used together with care and clarity, the future becomes a little more knowable.
