Every single system on the TOP500 list of the world’s most powerful supercomputers runs Linux. That 100% adoption rate has held since November 2017. This article breaks down how Linux dominates scientific computing across supercomputers, HPC clusters, AI research infrastructure, cloud-based research platforms, and major laboratories worldwide — with the actual numbers behind that dominance.
Top Linux In Scientific Computing Statistics (2026)
- 100% of TOP500 supercomputers run Linux, a streak now in its eighth consecutive year.
- 96.7% of scientific simulations — climate modeling, genomic analysis, particle physics — execute on Linux.
- 87.8% of machine learning workloads globally run on Linux infrastructure in 2025.
- Four systems on the TOP500 have crossed the exascale threshold, all running Linux-based operating systems.
- Google Cloud reports 91.6% Linux adoption, while AWS EC2 runs 83.5% Linux instances for research computing.
How Dominant Is Linux In TOP500 Supercomputers?
The November 2025 TOP500 list — the 66th edition — recorded 22.16 exaFLOPS of aggregate performance across all 500 systems. Four of those machines have crossed the exascale barrier. All four run Linux.
| Rank | System | Location | Performance (ExaFLOPS) | OS |
|---|---|---|---|---|
| 1 | El Capitan | Lawrence Livermore National Lab, USA | 1.809 | RHEL |
| 2 | Frontier | Oak Ridge National Lab, USA | 1.353 | Cray Linux |
| 3 | Aurora | Argonne National Lab, USA | 1.012 | Linux (Intel) |
| 4 | JUPITER Booster | Jülich Supercomputing Centre, Germany | 1.000 | Linux (Bull) |
| 5 | Eagle | Microsoft Azure Cloud, USA | 0.561 | Linux |
Source: TOP500.org, November 2025; HPCwire, November 2025
All three U.S. exascale systems sit in Department of Energy facilities. JUPITER, hosted at Forschungszentrum Jülich in Germany, became Europe’s first exascale system in November 2025. It had already powered over 100 research projects by that point, including Earth-system modeling at one-kilometer resolution.
The United States leads the TOP500 with 175 systems and 48.4% of total computing power as of June 2025. Linux runs on every one of them — and on every other system in the list, regardless of country.
Which Linux Distributions Do Research Labs Prefer?
Choices cluster around a handful of RHEL-based options. CERN and Fermilab jointly adopted AlmaLinux as their standard distribution in December 2022, replacing CentOS 7 and Scientific Linux 7 after both reached end-of-life in June 2024. NASA’s Pleiades supercomputer runs SUSE Linux Enterprise Server and is planned to move to Red Hat Enterprise Linux.
| Distribution | Enterprise Market Share (2025) | Primary Research Users |
|---|---|---|
| Red Hat Enterprise Linux | 43.1% | DOE National Labs, NASA |
| AlmaLinux | Growing (post-CentOS) | CERN, Fermilab, DESY, ETH Zurich |
| SUSE Linux Enterprise | 11.2% | NASA Pleiades, manufacturing HPC |
| Ubuntu | 33.9% (general), 27.7% (developer) | University research groups, ML developers |
| Cray Linux Environment | Custom (HPE systems) | Frontier, LUMI, Alps |
Source: CommandLinux.com, Dec 2025; Stack Overflow Developer Survey 2025; Neowin, Oct 2024
CERN’s Linux support portal confirms AlmaLinux 10.1 is in production as of late 2025, alongside RHEL 10.1 and 9.7. The Linux Foundation, which now manages over 1,000 active projects, provides the upstream kernel that all these distributions build on.
How Much Of Scientific HPC Runs On Linux?
Beyond the TOP500 headline, Linux dominates the broader HPC workloads that research teams depend on daily. Climate modeling, genomic sequencing, and particle physics simulations run on Linux in 96.7% of cases. The remaining 3.3% lands on legacy Unix or specialized real-time operating systems tied to specific instruments.
| Metric | Percentage |
|---|---|
| TOP500 supercomputers on Linux | 100% |
| Scientific simulations (climate, genomic) on Linux | 96.7% |
| ML workloads on Linux infrastructure | 87.8% |
| HPC institutions citing community support for Linux | 73.2% |
| Government HPC labs increasing Linux kernel customization (YoY) | 15.6% |
| Production Kubernetes clusters on Linux | 96.4% |
Source: SQ Magazine, November 2025; CNCF, 2025; TOP500.org
Government-funded HPC laboratories reported a 15.6% year-over-year increase in Linux kernel customization in 2025. These labs modify the kernel to extract performance gains matched to their specific hardware and workload profiles.
What Role Does Linux Play In AI And Machine Learning Research?
AI research is one of the fastest-growing segments of scientific computing, and Linux powers nearly all of it. Machine learning workloads showed 87.8% reliance on Linux globally in 2025. Around 78.5% of developers use Linux as a primary or secondary OS for AI and ML work.
| AI/ML Metric | Percentage |
|---|---|
| ML workloads on Linux infrastructure | 87.8% |
| Developers using Linux for AI/ML | 78.5% |
| GPU-accelerated systems’ share of TOP500 compute power | 86.2% |
| AI/ML workloads on Kubernetes | 54% |
| Deep learning papers using PyTorch | 85% |
| Docker adoption for ML workflows (up 17 pts YoY) | 71% |
Source: IndustryResearch.biz, 2025; CommandLinux.com, January 2026; CNCF, 2025
Accelerator-equipped systems — mainly GPU clusters running NVIDIA hardware on Linux — account for 86.2% of all computational power in the November 2025 TOP500. The count of accelerated systems grew from 194 to 211 between June and November 2024.
PyTorch, which became a Linux Foundation project in 2022, captured 55% of the production ML framework market in Q3 2025. Roughly 85% of deep learning research papers now reference PyTorch. The broader Linux job market reflects this trend, with AI/ML roles now among the fastest-growing categories for Linux professionals.
How Are Research Institutions Using Linux In The Cloud?
Cloud-based research computing has opened access for universities that can’t fund on-premises supercomputers. Linux is the default across all three major providers.
| Cloud Provider | Linux VM Percentage (2025) |
|---|---|
| Google Cloud | 91.6% |
| AWS EC2 | 83.5% |
| Microsoft Azure | 61.8% |
| All public cloud workloads (average) | 90% |
Source: SQ Magazine, November 2025; CommandLinux.com, December 2025
AWS announced a dedicated research HPC cluster with 40,000 Trainium chips for university labs in late 2024 — all running Linux. University HPC systems are growing at around 18% CAGR, while national lab systems expand at roughly 43% CAGR and industrial AI clusters at about 78% CAGR. All of these run Linux.
The Linux server market overall reached $21.97 billion in 2024 and continues to expand as research computing migrates to cloud and hybrid infrastructure.
How Big Is The Linux Market Tied To Scientific Computing?
The broader Linux OS market reflects strong demand from scientific and enterprise computing sectors.
| Year | Global Linux OS Market Value |
|---|---|
| 2024 | $21.97 billion |
| 2025 (estimated) | $26.41 billion |
| 2032 (projected) | $99.69 billion |
Source: XtendedView, December 2025; IndustryResearch.biz, 2025
Note: XtendedView publishes two market figures using different methodologies. A narrower segmentation shows $7.64 billion in 2024 growing to $10.73 billion in 2026 at 19% CAGR. The broader definition ($21.97 billion in 2024) includes bundled enterprise services and cloud platform revenue. The HPC market overall was valued at $44.2 billion in 2024 and is projected to reach $117.6 billion by 2034 at a 10.4% CAGR, with scientific research holding the largest application share.
Key Takeaways From Linux Adoption In Research
Linux controls 100% of the world’s top supercomputers — a position held for eight straight years. Labs like CERN and Fermilab have standardized on RHEL-based distributions, while cloud research computing runs 83–92% Linux depending on the provider. Scientific simulations use Linux in 96.7% of workloads, and 87.8% of ML infrastructure is Linux-based.
For research institutions evaluating their stack, the question isn’t whether to use Linux. It’s which distribution fits their hardware, support needs, and workload profile. Country-level Linux adoption data shows the same pattern playing out beyond research — governments and enterprises are standardizing on Linux for the same reasons labs did years ago.
FAQs
What percentage of supercomputers run Linux?
100% of the TOP500 supercomputers run Linux. This has been the case since November 2017, when the last two non-Linux systems dropped off the list.
Which Linux distribution does CERN use?
CERN adopted AlmaLinux as its standard distribution in December 2022, replacing CentOS 7. It also runs RHEL in parallel, with AlmaLinux 10.1 and RHEL 10.1 both in production as of late 2025.
How much of AI and machine learning research uses Linux?
87.8% of machine learning workloads run on Linux infrastructure globally. About 78.5% of developers working in AI/ML use Linux as their primary or secondary operating system.
What is the global Linux operating system market size?
The global Linux OS market reached $21.97 billion in 2024. Projections estimate it will grow to $99.69 billion by 2032 at a 20.9% compound annual growth rate.
Why do research institutions prefer Linux over other operating systems?
Linux offers kernel-level customization for specific hardware, zero licensing costs at scale, and strong community support. Government HPC labs increased kernel customization by 15.6% year-over-year in 2025 to optimize performance.
