Dock Line
Researchers stand before a large curved wall of display screens in the Texas Advanced Computing Center Visualization Lab, photographed June 24, 2013
The Texas Advanced Computing Center Visualization Lab, photographed June 24, 2013. This is not a picture of the new Horizon machine. A Nov. 17 UT Austin release says Horizon, TACC's newest supercomputer, has opened to early users. (UNM Research Data Services)

Jane Long

November 18, 2026 · 4 min read

UT Austin's Horizon Supercomputer Opens to Early Users

A Nov. 17 University of Texas at Austin release says the NSF-funded Horizon supercomputer at the Texas Advanced Computing Center has opened to early users and ranks No. 24 on the November 2026 TOP500 list.

The Texas Advanced Computing Center at The University of Texas at Austin has opened its newest machine, Horizon, to early users, according to a university news release dated Nov. 17. Horizon is a 4,000-graphics-processing-unit system and the centerpiece of the National Science Foundation's Leadership-Class Computing Facility.

TACC says the center remains home to the nation's most powerful academic supercomputer. On the November 2026 TOP500 list of the world's fastest machines, which Dock Line reviewed, Horizon ranks No. 24, with a measured High-Performance Linpack result of 132 petaflops and a peak rating of 167.14 petaflops on 144,000 cores. Those figures describe current scientific computing speed, not a finished project. The university says Horizon's performance will climb further when CPU capacity is installed in early 2027.

"With Horizon, NSF is making its biggest-ever investment in computing infrastructure, and we can't wait to see the discoveries it makes possible," said Katie Antypas, senior science and engineering adviser for the NSF Computer and Information Science and Engineering Directorate, in the release. "After years of planning and building, it's exciting to finally put this system into the hands of researchers."

What the Machine Is Built to Do

The university says that with Horizon, TACC has amassed nearly 5,500 GPUs, which it calls the highest concentration of computing power at any university in the United States. Horizon was developed with NVIDIA, Dell Technologies, and VAST Data. The first phase uses 4,000 GPUs on NVIDIA's Grace Blackwell platform. A second phase, planned for early 2027, is to add NVIDIA Vera CPU servers.

TACC executive director Dan Stanzione, who is also associate vice president for research at UT Austin, called the machine "an engine for discovery, a balanced system for both AI and simulation." The release lists intended uses that mix artificial intelligence, numerical simulation, and inference, and it names fields including medicine, earthquake science, quantum materials, lattice quantum chromodynamics, biophysics, and astronomy.

Those are aims, not results already in hand. Fernanda Leite, UT Austin's vice president for research, said in the release that Horizon is "built for questions that matter most to Texas and the nation but are too big, too complex or too risky to test in a lab." She offered examples: tracing how a flash flood would move through a Hill Country watershed, stress-testing the power grid before the next winter storm, or analyzing biological data so a treatment might reach patients sooner. None of those studies is reported as complete in the announcement.

The university's specification sheet is ambitious. It lists 140 GPU peak petaflops, with 180 petaflops projected once CPUs are added, and it claims a tenfold improvement in simulation speed over Frontera, TACC's previous flagship. That "10 times" figure is TACC's claim for simulation work, not a direct comparison of the public Linpack scores. On the TOP500 list, Frontera is still recorded at No. 96, at 23.52 petaflops, even though the university says that machine was retired in October. The release also cites 20 exaflops of AI performance at BF16/FP16 precision and 80 exaflops at FP4, local all-solid-state storage of 400 petabytes, and NVIDIA Quantum-X800 InfiniBand networking. Dock Line has not independently benchmarked those AI and storage figures.

The system is powered by a new 20-megawatt data center with liquid cooling in Round Rock, which the university says is up to six times more energy efficient than prior systems. A 20-megawatt load is still a serious draw on the electric grid, and Texans have reason to watch how large computing sites use power even when the work is scientific rather than commercial.

From Frontera to Early Users

TACC retired Frontera in October after more than seven years of service. The machine entered production in June 2019, became the longest-running system in TACC's history, and completed more than 8 million jobs, the release says. The university credits earlier TACC systems with work by more than 130,000 students and researchers, and it points to research ranging from building-code changes meant to improve hurricane resilience to work connected with the 2026 Nobel Prize in physics awarded to Francis Halzen for contributions to the NSF IceCube Neutrino Observatory. Those credits come from the university release; Dock Line has not independently verified the Nobel connection.

Vista, TACC's AI-focused machine built on 600 NVIDIA Grace Hopper Superchips, served as the prototype for Horizon and was the first TACC system in nearly 20 years to move away from x86 processors, according to the release. With Frontera retired, TACC's production fleet now includes Horizon, Vista, Lonestar6, Stampede3, Jetstream2, Ranch, and Corral.

John West, deputy director of the NSF Leadership-Class Computing Facility, said Horizon is "only part of the story," and that the facility is also building software, storage, user support, training, and partnerships "that will help researchers turn its power into discovery." Construction of the facility is supported by NSF Award No. 2323116.

Early-user access is the start of the work, not the finish. The CPUs are still months away, the discoveries TACC describes have not yet been delivered, and the public ranking already places Horizon well below the largest government and industrial machines. What Texas has, as of this week, is a federally funded academic system now open to a first group of researchers, with a second phase still to come.