Aurora
An exascale supercomputer at Argonne National Laboratory built by Intel and Cray.
Aurora is an exascale supercomputer at Argonne National Laboratory, sponsored by the U.S. Department of Energy and designed by Intel and Cray. Fully installed in June 2023, it reached 1.012 exaFLOPS on the TOP500 list and supports open scientific research across fields from fusion to cosmology.
- Fully installed on 22 June 2023 at Argonne National Laboratory
- Measured performance of 1.012 exaFLOPS on the TOP500 list
- 10,624 nodes with Intel Xeon Max CPUs and Intel Max GPUs
- About 10 petabytes of memory and 230 petabytes of storage
- Estimated cost of US$500 million as projected in 2019
What it is
Aurora is a large research supercomputer housed at the Argonne Leadership Computing Facility and operated for the U.S. Department of Energy. Intel and Cray designed the machine for Argonne National Laboratory. It is built from 10,624 nodes, each pairing two Intel Xeon Max processors with six Intel Max series GPUs, giving roughly 130 teraFLOPS per node. Total memory is around 10 petabytes, with 230 petabytes of storage. Measured performance stands at 1.012 exaFLOPS, with a theoretical peak of 1.98 exaFLOPS. The system draws about 38.7 megawatts and cost an estimated US$500 million as projected in 2019. Its purpose is open scientific research and development.
How it came about
The Department of Energy set out a plan in 2013 for a machine reaching about one exaFLOP within a 20 megawatt power budget by 2020. Aurora was revealed in 2015 with a target completion of 2018 and an expected speed near 180 petaFLOPS, close to Summit. Intel said in 2017 that delivery would slip to 2021 while the design grew to one exaFLOP. A further six-month delay was announced in October 2020, costing it the title of first US exascale system. Intel claimed in late October 2021 that peak double-precision output would pass two exaFLOPS, a figure never achieved. Installation finished on 22 June 2023.
The compute
Each of the 10,624 nodes carries two Intel Xeon Max CPUs alongside six Intel Max series GPUs in a unified memory layout, delivering up to 130 teraFLOPS per node. Aggregate memory is about 10 petabytes and storage capacity reaches 230 petabytes. The whole installation draws close to 39 megawatts. For context, El Capitan runs on 30 megawatts and Frontier on 24 megawatts. Peak double-precision throughput is rated at 1.98 exaFLOPS while measured performance is 1.012 exaFLOPS.
Where the GPUs are
The machine sits inside the Argonne Leadership Computing Facility at Argonne National Laboratory. It is run jointly by the laboratory and the Department of Energy. Power consumption for the installation is stated at roughly 38.7 megawatts.
Customers and deals
Aurora is intended for use by the wider scientific community. Workloads include studies of brain structure, nuclear fusion, low-carbon technologies, subatomic particles, cancer and cosmology, plus design of new materials for batteries and more efficient solar cells.
Money and ownership
A 2019 estimate put the cost at US$500 million. The project was sponsored by the Department of Energy, with Intel and Cray responsible for the design.
Its place in AI
Aurora briefly held second place on the TOP500 ranking of the world's fastest computers between November 2023 and June 2024. In May 2024 it appeared at number two with 1.012 exaFLOPS, becoming the second exascale-class system on that list. As of June 2026 it stands at number four. Olivier Franza serves as chief architect and principal investigator.
Questions and criticism
Repeated schedule slips pushed delivery past 2018, 2021 and a further six-month extension, so Aurora lost the chance to be the first exascale machine in the United States. Intel's late-2021 projection of more than two exaFLOPS peak was never realised. Power draw of nearly 39 megawatts exceeds that of comparable systems such as El Capitan and Frontier.
Questions people ask about Aurora
What is Aurora?
Aurora is an exascale supercomputer located at the Argonne Leadership Computing Facility. The U.S. Department of Energy sponsored it, and Intel together with Cray designed it for Argonne National Laboratory. It entered full installation in June 2023 and serves open scientific research.
How fast is it?
Measured performance is 1.012 exaFLOPS, with a theoretical peak of 1.98 exaFLOPS. On the TOP500 ranking it reached second place in May 2024 and stood at fourth place as of June 2026.
What hardware does it use?
There are 10,624 nodes. Every node combines two Intel Xeon Max processors with six Intel Max series GPUs in a unified memory design, giving up to 130 teraFLOPS per node. Memory totals around 10 petabytes and storage totals 230 petabytes.
What research runs on it?
Scientists use it for work on brain structure, nuclear fusion, low-carbon technology, subatomic particles, cancer and cosmology. It also supports material design for batteries and improved solar cells, and is open to the general research community.
How much power does it draw?
The installation consumes about 38.7 megawatts. That is more than El Capitan at 30 megawatts and Frontier at 24 megawatts, both of which appear on the TOP500 list.
Good reading and sources
- Official websiteOfficial site
- America鈥檚 first exascale supercomputer to be built by 2021The Verge, Article
- Checking your browserAnandTech, Article
- Intel NewsroomIntel, Article
The facts on this page come from the Wikipedia article "Aurora (supercomputer)", written by its contributors and used under the CC BY-SA 4.0 licence, and from its Wikidata record (CC0). EveryNeocloud rewrote the text in its own words, with AI assistance, and compares every page with the article to make sure no passage is copied. It may be reused under the same CC BY-SA 4.0 licence with a link back.
It may contain mistakes, and figures change over time. Nothing here is investment or buying advice. If something is wrong, tell us. Last refreshed 2026-11-14.