News
DealsCorvex agrees USD 103 million take-or-pay deal to perform AI inference with Mid-Atlantic GPUsSovereignNeysa scales AI infrastructure in India with AMD Instinct MI350X GPUs on TytanInferenceMicrosoft Foundry adds Baseten, DeepSeek AI inference (MSFT:NASDAQ)ClustersDataOne and Rowan University sign engineering co-op program agreement for DataOne's Vineland AI factoryClustersLook inside Cal Poly's $3 million AI Factory (Video)DealsOpenAI Taps AMD Chips for 457 MW of AI Capacity in Six-Year TensorWave DealClustersCal Poly launches an AI data center on campus. Take a look insideInferenceArista takes aim at the next AI network battleground: distributed inferenceDealsOpenAI Signs 457 MW AI Data Center Deal as AMD Expands Its Challenge to NvidiaClustersBehind Dell AI Factory鈥檚 NVIDIA & AMD Enterprise ExpansionDealsCorvex agrees USD 103 million take-or-pay deal to perform AI inference with Mid-Atlantic GPUsSovereignNeysa scales AI infrastructure in India with AMD Instinct MI350X GPUs on TytanInferenceMicrosoft Foundry adds Baseten, DeepSeek AI inference (MSFT:NASDAQ)ClustersDataOne and Rowan University sign engineering co-op program agreement for DataOne's Vineland AI factoryClustersLook inside Cal Poly's $3 million AI Factory (Video)DealsOpenAI Taps AMD Chips for 457 MW of AI Capacity in Six-Year TensorWave DealClustersCal Poly launches an AI data center on campus. Take a look insideInferenceArista takes aim at the next AI network battleground: distributed inferenceDealsOpenAI Signs 457 MW AI Data Center Deal as AMD Expands Its Challenge to NvidiaClustersBehind Dell AI Factory鈥檚 NVIDIA & AMD Enterprise Expansion
EveryNeocloud
El Capitan

El Capitan

Exascale supercomputer for US nuclear stockpile stewardship built by HPE and AMD.

El Capitan is an exascale supercomputer located at Lawrence Livermore National Laboratory in California. It became operational in 2024 to support the US Department of Energy's nuclear weapons programme.

Share on WhatsApp

  • Peak performance reaches 2.821 exaFLOPS.
  • Uses 43,808 AMD Instinct MI300A GPUs.
  • Cost estimated at $600 million.
  • Dedicated January 9, 2025.
  • World's fastest from Nov 2024 to Jun 2026.

What it is

El Capitan is a powerful exascale supercomputer hosted at the Lawrence Livermore National Laboratory in Livermore, California. Built on the Hewlett Packard Enterprise Cray EX Shasta architecture, it serves as the third exascale system deployed by the United States. The machine operates with a peak performance of 2.821 exaFLOPS and a sustained rate of 1.809 exaFLOPS. Its primary function is to support scientific research and the stockpile stewardship programme managed by the National Nuclear Security Administration. The system was officially dedicated in early 2025 after becoming operational the previous year.

How it came about

The project originated from the Department of Energy's CORAL-2 initiative, designed to replace the earlier Sierra supercomputer. Lawrence Livermore National Laboratory collaborated with Hewlett Packard Enterprise and Advanced Micro Devices to construct the facility. Initial prototypes achieved significant benchmarks before full deployment began in mid-2023. By November 2024, TOP500 verified El Capitan as the world's fastest supercomputer, displacing Frontier. It retained this title until June 2026 when it was surpassed by LineShine. The official dedication ceremony occurred on January 9, 2025, attended by industry leaders.

The compute

The system utilises 43,808 AMD Instinct MI300A accelerators combined with AMD fourth-generation EPYC CPUs. These components share a coherent on-chip interconnect within a single organic package. Each accelerator features 24 Zen4 CPU cores and CDNA3-based GPU elements integrated with 128GB of HBM3 memory. Interconnectivity relies on HPE Slingshot switches providing 12.8 terabits per second of bandwidth. Nodes are arranged in a dragonfly topology using optical or copper cabling spanning 145 kilometres. The entire installation occupies approximately 7,500 square feet of floor space within liquid-cooled racks.

Where the GPUs are

El Capitan resides at the Livermore Computing Complex in Livermore, California. The physical infrastructure requires substantial power, drawing up to 30 megawatts. The computing environment consists of at least 87 compute racks housing the processing nodes and storage arrays. The design prioritises minimal cable length through customised cabling solutions. This location supports sensitive and classified tasks related to national security. The facility represents a significant investment in domestic high-performance computing capabilities.

Customers and deals

The primary customer is the United States government, specifically the National Nuclear Security Administration. The machine supports the stockpile stewardship programme, which involves maintaining the safety and reliability of the US nuclear arsenal. Scientific research and development constitute additional use cases for the system. The collaboration between the laboratory and technology providers ensures the hardware meets stringent government requirements. No external commercial customers are mentioned in the available records.

Money and ownership

Construction costs are estimated at $600 million. The project was funded through the Department of Energy's CORAL-2 initiative. Financial details regarding specific funding rounds or revenue generation are not disclosed. The investment reflects the high expense associated with exascale computing infrastructure. Government sponsorship drives the procurement process rather than market competition. The economic impact focuses on technological advancement and national security rather than direct profit.

Its place in AI

El Capitan held the number two position in the TOP500 rankings as of June 2026. It displaced Frontier at Oak Ridge National Laboratory to claim the top spot in November 2024. The system demonstrates the capability of AMD and HPE technologies in extreme performance scenarios. Industry executives highlighted parallels between this supercomputer and artificial intelligence training infrastructure. Its existence underscores the strategic importance of domestic supercomputing power for national defence objectives.

Questions and criticism

The system consumes significant energy, requiring 30 megawatts of power. Export controls may restrict the sharing of certain high-performance computing technologies. The concentration of such advanced capabilities within government facilities raises questions about accessibility. Environmental impacts associated with high-power data centres remain a consideration. The reliance on specific proprietary architectures could limit future flexibility. These factors represent standard challenges in large-scale supercomputing deployments.

Questions people ask about El Capitan

What is El Capitan used for?

El Capitan primarily supports the stockpile stewardship programme of the US National Nuclear Security Administration. It also facilitates general scientific research and development. The system handles sensitive and classified tasks related to the maintenance of the US nuclear weapons arsenal.

Who built El Capitan?

The supercomputer was built through a partnership between Lawrence Livermore National Laboratory, Hewlett Packard Enterprise, and Advanced Micro Devices. The Department of Energy sponsored the project under its CORAL-2 initiative to replace the older Sierra system.

How powerful is El Capitan?

El Capitan delivers a maximum sustained performance of 1.809 exaFLOPS and a peak performance of 2.821 exaFLOPS. It utilises over eleven million CPU and GPU cores across its architecture. This made it the world's fastest supercomputer from late 2024 until mid-2026.

Good reading and sources

Where this page comes from

The facts on this page come from the Wikipedia article "El Capitan (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 2026-11-15.

Photo credits on this page

The small pictures above come from Wikimedia Commons under free licences. Some are cropped to fit.

The weekly digest

Every Monday: the provider of the day, new pages on EveryNeocloud, and the conferences and anniversaries of the week ahead.