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 TytanSovereignNeysa Deploys AMD GPU Cluster in Mumbai for AI Training and InferenceInferenceMicrosoft 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 NvidiaDealsCorvex 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 TytanSovereignNeysa Deploys AMD GPU Cluster in Mumbai for AI Training and InferenceInferenceMicrosoft 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 Nvidia
EveryNeocloud
Horizon

Horizon

Largest academic supercomputer in the United States

Horizon is the largest academic supercomputer in the United States, operated by the Texas Advanced Computing Center at The University of Texas at Austin. Funded by the National Science Foundation, it uses Nvidia Grace Blackwell GPUs to deliver exascale AI performance for scientific research.

Share on WhatsApp

  • Largest academic supercomputer in the United States
  • 4,000 Nvidia Grace Blackwell GPUs
  • 80 exaFLOPS AI performance at FP4 precision
  • Located in Round Rock, Texas with liquid cooling
  • Supports 10,000 to 15,000 researchers annually

What it is

Horizon is a leadership-class supercomputer designed for open science and artificial intelligence research. It is operated by the Texas Advanced Computing Center (TACC) at The University of Texas at Austin. As the centerpiece of the National Science Foundation鈥檚 Leadership-Class Computing Facility, it provides unprecedented computational power to U.S. researchers. The system features 4,000 Nvidia GPUs and aims to support breakthroughs in fields such as climate science, medicine, and physics.

How it came about

The National Science Foundation launched the installation of Horizon in late 2025. It represents the largest investment the agency has made in computing infrastructure. The project was developed in collaboration with industry partners including Dell Technologies, Nvidia, and VAST Data. Phase 1 of the system is scheduled to enter production in Spring or Fall 2026, with Phase 2 adding Nvidia Vera CPU servers in early 2027. It succeeds Frontera as the top academic supercomputer in the country.

The compute

The system runs on the Nvidia Grace Blackwell platform, featuring 4,000 GPUs and one million CPU cores. It delivers 20 exaFLOPS of AI performance at BF16 or FP16 precision, and 80 exaFLOPS at FP4. This represents a more than 100-fold improvement over previous systems for AI workloads. The nodes are interconnected using Nvidia Quantum-X800 InfiniBand networking with in-network computing capabilities. Storage includes 400 petabytes of local all-solid state storage with high bandwidth.

Where the GPUs are

Horizon is housed in a new data centre in Round Rock, Texas, provided by Sabey Data Centers. The facility uses advanced liquid cooling to manage heat from high-density deployments. It operates with a power capacity between 15 and 20 megawatts. The site is designed to be up to six times more energy efficient than prior generations. Alongside Horizon, the Ranch archive system provides exabyte-scale storage for research data.

Customers and deals

The facility serves the national research community, supporting between 10,000 and 15,000 users annually. Access is granted through a peer-review process to students, faculty, national laboratories, nonprofits, and industry partners working on NSF-funded projects. International collaborators on these projects may also use the resource. The Characteristic Science Applications programme involves eleven projects that help shape the system design.

Money and ownership

The project is funded by the National Science Foundation under Award #2323116. It is described as the largest investment the NSF has made in computing infrastructure. The system was developed in partnership with commercial technology providers such as Dell, Nvidia, and VAST Data, though specific financial terms of these collaborations are not detailed in the public materials.

Its place in AI

Horizon is positioned as a national instrument for discovery, comparable to major scientific initiatives like the James Webb Space Telescope. It aims to redefine possibilities in scientific computing by combining massive parallelism with AI-driven data contextualisation. The system supports the Characteristic Science Applications programme, ensuring real-world scientific codes are tuned for the new architecture before general availability.

Questions people ask about Horizon

Who can use the Horizon supercomputer?

Access is open to U.S. researchers, engineers, students, and international collaborators on NSF-funded projects. Resources are allocated through a rigorous peer-review process to ensure they support innovative and impactful research across various scientific disciplines.

What hardware powers the Horizon system?

Horizon uses the Nvidia Grace Blackwell platform with 4,000 GPUs and one million CPU cores. It features Nvidia Quantum-X800 InfiniBand networking for high-speed interconnectivity and 400 petabytes of local solid-state storage for rapid data access.

When will Horizon be available for research?

Phase 1 of the system is scheduled to enter production in 2026. Phase 2, which includes additional Nvidia Vera CPU servers, is planned for early 2027. The installation process began in late 2025 at the Texas Advanced Computing Center.

Good reading and sources

    Where this page comes from

    There is no English Wikipedia article on Horizon yet, so the facts on this page come from the sources listed above: tacc.utexas.edu/systems/horizon/, tacc.utexas.edu/news/latest-news/2025/11/17/nsf-lccf-horizon-supercomp. Some are the subject's own pages, so its claims about itself are reported as its claims. EveryNeocloud wrote the text in its own words, with AI assistance, and compares it with the sources to make sure no passage is copied.

    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.

    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.