COMPUTE ATLAS Supercomputer supply-chain graph
1540 systems 545 sites Sourced data

Supercomputers/Analysis/Interconnect share

Analysis 02

Interconnect share over time

The fabric layer consolidated twice in twelve months and almost nobody charted it. NVIDIA completed its acquisition of Mellanox in April 2020; HPE completed its acquisition of Cray in September 2019. Four independent fabric suppliers became two corporate owners, and because a spec string names the product rather than the parent, the change is invisible unless you resolve the entities first. As of 2026, Hewlett Packard Enterprise accounts for 43% of installed FLOPS by fabric and NVIDIA for 38%.

Researched with AI assistance, checked by scripts. These rows were researched by AI agents reading public operator, laboratory and vendor sources, appended by scripts and checked by automated validation; no subject-matter expert has reviewed them (how this site is made). A mechanical second-reader pass has since fetched every cited source: 1350 of 1540 systems have a readable citation that names them, and 64 are genuinely weakly sourced. Every claim carries its source and a confidence tier. Treat anything below verified as a lead, not a citation.

Read the shape, not the spikes. This is the installed base of the 1540 systems in this dataset: a curated seed chosen for supply-chain diversity, not a census of the market. At that sample size one machine entering or leaving service moves a share line by tens of points, so year-to-year jumps are composition effects rather than market events. The multi-year trends are the part worth quoting; the individual steps usually have one system's name on them, and the table view under the chart will tell you which.

  • Hewlett Packard Enterprise
  • NVIDIA
  • Shenzhen Cloud Computing Center
  • Fujitsu
  • Broadcom
  • Intel
  • Atos
  • Other (17 suppliers)
Interconnect share of installed FLOPS, 1993–2026 Stacked area chart of each interconnect supplier's share of installed Rmax by year, with suppliers rolled up to the corporate parent that owns them today. Hewlett Packard EnterpriseNVIDIAShenzhen Cloud Computing CenterFujitsuBroadcomIntelAtosOther (17 suppliers) 0%25%50%75%100%19931998200320082013201820232026 HPENVIDIAShenzhen Cloud Computing Center 1993 HPE 0.0% NVIDIA 0.0% Shenzhen Cloud Computing Center 0.0% Fujitsu 29.3% Broadcom 0.0% Intel 37.5% Atos 0.0% Other 33.2%1994 HPE 15.2% NVIDIA 0.0% Shenzhen Cloud Computing Center 0.0% Fujitsu 24.8% Broadcom 0.0% Intel 31.8% Atos 0.0% Other 28.2%1995 HPE 10.5% NVIDIA 0.0% Shenzhen Cloud Computing Center 0.0% Fujitsu 30.7% Broadcom 0.0% Intel 39.4% Atos 0.0% Other 19.4%1996 HPE 32.5% NVIDIA 0.0% Shenzhen Cloud Computing Center 0.0% Fujitsu 20.0% Broadcom 0.0% Intel 12.3% Atos 0.0% Other 35.1%1997 HPE 36.9% NVIDIA 0.0% Shenzhen Cloud Computing Center 0.0% Fujitsu 12.9% Broadcom 0.0% Intel 30.5% Atos 0.0% Other 19.7%1998 HPE 44.2% NVIDIA 0.0% Shenzhen Cloud Computing Center 0.0% Fujitsu 6.5% Broadcom 0.0% Intel 15.3% Atos 0.0% Other 34.1%1999 HPE 45.8% NVIDIA 0.0% Shenzhen Cloud Computing Center 0.0% Fujitsu 5.1% Broadcom 0.0% Intel 12.1% Atos 0.0% Other 36.9%2000 HPE 27.0% NVIDIA 0.0% Shenzhen Cloud Computing Center 0.0% Fujitsu 7.7% Broadcom 0.0% Intel 7.1% Atos 0.0% Other 58.2%2001 HPE 15.2% NVIDIA 0.0% Shenzhen Cloud Computing Center 0.0% Fujitsu 4.0% Broadcom 0.0% Intel 4.0% Atos 0.0% Other 76.7%2002 HPE 4.7% NVIDIA 0.0% Shenzhen Cloud Computing Center 0.0% Fujitsu 1.2% Broadcom 0.0% Intel 1.2% Atos 0.0% Other 92.8%2003 HPE 12.9% NVIDIA 6.6% Shenzhen Cloud Computing Center 0.0% Fujitsu 0.8% Broadcom 0.0% Intel 0.9% Atos 0.0% Other 78.9%2004 HPE 2.9% NVIDIA 1.5% Shenzhen Cloud Computing Center 0.0% Fujitsu 0.2% Broadcom 0.0% Intel 0.2% Atos 0.0% Other 95.2%2005 HPE 5.9% NVIDIA 1.1% Shenzhen Cloud Computing Center 0.0% Fujitsu 0.1% Broadcom 0.0% Intel 0.2% Atos 0.0% Other 92.7%2006 HPE 5.0% NVIDIA 1.6% Shenzhen Cloud Computing Center 0.0% Fujitsu 1.1% Broadcom 0.0% Intel 0.1% Atos 0.0% Other 92.2%2007 HPE 39.8% NVIDIA 3.8% Shenzhen Cloud Computing Center 0.0% Fujitsu 0.5% Broadcom 0.0% Intel 0.0% Atos 0.0% Other 55.9%2008 HPE 12.3% NVIDIA 66.9% Shenzhen Cloud Computing Center 0.0% Fujitsu 0.1% Broadcom 0.0% Intel 0.0% Atos 0.0% Other 20.6%2009 HPE 21.1% NVIDIA 62.8% Shenzhen Cloud Computing Center 0.0% Fujitsu 0.1% Broadcom 0.0% Intel 0.0% Atos 0.0% Other 16.0%2010 HPE 27.7% NVIDIA 51.4% Shenzhen Cloud Computing Center 0.0% Fujitsu 0.0% Broadcom 0.0% Intel 0.0% Atos 0.0% Other 20.9%2011 HPE 21.2% NVIDIA 39.2% Shenzhen Cloud Computing Center 0.0% Fujitsu 23.2% Broadcom 0.0% Intel 3.6% Atos 0.0% Other 12.8%2012 HPE 22.9% NVIDIA 27.9% Shenzhen Cloud Computing Center 0.0% Fujitsu 9.3% Broadcom 0.0% Intel 1.3% Atos 0.0% Other 38.6%2013 HPE 14.9% NVIDIA 26.5% Shenzhen Cloud Computing Center 0.0% Fujitsu 5.5% Broadcom 0.0% Intel 0.8% Atos 0.0% Other 52.3%2014 HPE 16.1% NVIDIA 31.7% Shenzhen Cloud Computing Center 0.0% Fujitsu 4.8% Broadcom 0.0% Intel 0.9% Atos 0.0% Other 46.4%2015 HPE 21.4% NVIDIA 35.3% Shenzhen Cloud Computing Center 0.0% Fujitsu 6.7% Broadcom 0.0% Intel 1.0% Atos 0.0% Other 35.5%2016 HPE 19.4% NVIDIA 28.1% Shenzhen Cloud Computing Center 0.0% Fujitsu 4.2% Broadcom 0.0% Intel 6.2% Atos 0.0% Other 42.1%2017 HPE 23.9% NVIDIA 28.8% Shenzhen Cloud Computing Center 0.0% Fujitsu 3.2% Broadcom 0.0% Intel 9.6% Atos 2.0% Other 32.4%2018 HPE 15.9% NVIDIA 48.2% Shenzhen Cloud Computing Center 0.0% Fujitsu 2.0% Broadcom 0.0% Intel 10.7% Atos 1.2% Other 22.0%2019 HPE 14.2% NVIDIA 51.6% Shenzhen Cloud Computing Center 0.0% Fujitsu 1.8% Broadcom 0.0% Intel 11.6% Atos 1.1% Other 19.7%2020 HPE 8.5% NVIDIA 47.9% Shenzhen Cloud Computing Center 0.0% Fujitsu 23.9% Broadcom 0.0% Intel 7.0% Atos 0.7% Other 12.1%2021 HPE 12.7% NVIDIA 52.2% Shenzhen Cloud Computing Center 0.0% Fujitsu 19.4% Broadcom 0.0% Intel 5.5% Atos 1.4% Other 8.8%2022 HPE 46.2% NVIDIA 34.9% Shenzhen Cloud Computing Center 0.0% Fujitsu 10.4% Broadcom 0.0% Intel 2.9% Atos 0.8% Other 4.8%2023 HPE 46.5% NVIDIA 40.6% Shenzhen Cloud Computing Center 0.0% Fujitsu 7.3% Broadcom 0.0% Intel 1.6% Atos 0.5% Other 3.4%2024 HPE 55.7% NVIDIA 35.4% Shenzhen Cloud Computing Center 0.0% Fujitsu 4.6% Broadcom 0.0% Intel 1.0% Atos 1.1% Other 2.2%2025 HPE 48.5% NVIDIA 43.2% Shenzhen Cloud Computing Center 0.0% Fujitsu 3.6% Broadcom 1.2% Intel 0.6% Atos 0.9% Other 1.9%2026 HPE 43.2% NVIDIA 37.8% Shenzhen Cloud Computing Center 12.5% Fujitsu 2.9% Broadcom 1.0% Intel 0.5% Atos 0.7% Other 1.5%
Share of installed Rmax by fabric supplier. Mellanox counts toward NVIDIA from 2020 and Cray toward HPE from 2019, because the question is who owns the business now.
Download this chart's data (CSV) ↓
Table view: Interconnect share of installed FLOPS
YearHewlett Packard EnterpriseNVIDIAShenzhen Cloud Computing CenterFujitsuBroadcomIntelAtosOther (17 suppliers)Installed total
19930.0%0.0%0.0%29.3%0.0%37.5%0.0%33.2%423.1 GFlop/s
199415.2%0.0%0.0%24.8%0.0%31.8%0.0%28.2%499.0 GFlop/s
199510.5%0.0%0.0%30.7%0.0%39.4%0.0%19.4%725.0 GFlop/s
199632.5%0.0%0.0%20.0%0.0%12.3%0.0%35.1%2.31 TFlop/s
199736.9%0.0%0.0%12.9%0.0%30.5%0.0%19.7%4.44 TFlop/s
199844.2%0.0%0.0%6.5%0.0%15.3%0.0%34.1%8.86 TFlop/s
199945.8%0.0%0.0%5.1%0.0%12.1%0.0%36.9%11.16 TFlop/s
200027.0%0.0%0.0%7.7%0.0%7.1%0.0%58.2%18.95 TFlop/s
200115.2%0.0%0.0%4.0%0.0%4.0%0.0%76.7%33.65 TFlop/s
20024.7%0.0%0.0%1.2%0.0%1.2%0.0%92.8%108.96 TFlop/s
200312.9%6.6%0.0%0.8%0.0%0.9%0.0%78.9%156.01 TFlop/s
20042.9%1.5%0.0%0.2%0.0%0.2%0.0%95.2%687.37 TFlop/s
20055.9%1.1%0.0%0.1%0.0%0.2%0.0%92.7%894.12 TFlop/s
20065.0%1.6%0.0%1.1%0.0%0.1%0.0%92.2%1.06 PFlop/s
200739.8%3.8%0.0%0.5%0.0%0.0%0.0%55.9%2.25 PFlop/s
200812.3%66.9%0.0%0.1%0.0%0.0%0.0%20.6%11.18 PFlop/s
200921.1%62.8%0.0%0.1%0.0%0.0%0.0%16.0%16.24 PFlop/s
201027.7%51.4%0.0%0.0%0.0%0.0%0.0%20.9%26.70 PFlop/s
201121.2%39.2%0.0%23.2%0.0%3.6%0.0%12.8%45.36 PFlop/s
201222.9%27.9%0.0%9.3%0.0%1.3%0.0%38.6%125.4 PFlop/s
201314.9%26.5%0.0%5.5%0.0%0.8%0.0%52.3%212.5 PFlop/s
201416.1%31.7%0.0%4.8%0.0%0.9%0.0%46.4%242.9 PFlop/s
201521.4%35.3%0.0%6.7%0.0%1.0%0.0%35.5%316.0 PFlop/s
201619.4%28.1%0.0%4.2%0.0%6.2%0.0%42.1%501.7 PFlop/s
201723.9%28.8%0.0%3.2%0.0%9.6%2.0%32.4%660.7 PFlop/s
201815.9%48.2%0.0%2.0%0.0%10.7%1.2%22.0%1.08 EFlop/s
201914.2%51.6%0.0%1.8%0.0%11.6%1.1%19.7%1.20 EFlop/s
20208.5%47.9%0.0%23.9%0.0%7.0%0.7%12.1%2.00 EFlop/s
202112.7%52.2%0.0%19.4%0.0%5.5%1.4%8.8%2.54 EFlop/s
202246.2%34.9%0.0%10.4%0.0%2.9%0.8%4.8%4.73 EFlop/s
202346.5%40.6%0.0%7.3%0.0%1.6%0.5%3.4%7.08 EFlop/s
202455.7%35.4%0.0%4.6%0.0%1.0%1.1%2.2%11.20 EFlop/s
202548.5%43.2%0.0%3.6%1.2%0.6%0.9%1.9%14.20 EFlop/s
202643.2%37.8%12.5%2.9%1.0%0.5%0.7%1.5%17.65 EFlop/s
How this was calculated

Identical construction to the accelerator chart: each in-service system contributes its last reported Rmax to every year it was live, split evenly where a system has more than one fabric supplier.

Every system in this dataset has at least one interconnect edge, so unlike the accelerator chart there is no “none” band: a machine without a fabric is not a machine.

The rollup is the whole point of this chart. Suppliers are recorded as the company that shipped the fabric at build time and then resolved through vendor_aliases to today's parent. Read without that step, Summit's dual-rail EDR belongs to a company that no longer exists independently, and the concentration this chart shows does not appear.

One caveat that cuts against the chart: vertically integrated systems understate their supplier here. IBM's BlueGene fabrics and Fujitsu's Tofu are on the processor package, and appear in no spec string as a third-party product. They are recorded as IBM and Fujitsu edges, but a dataset built only from fabric-vendor mentions would miss them entirely.

Installed base of the 1540 systems in this dataset, not a census of the market.

The read

Three structural facts sit in this chart. The first is that the fabric layer has always been more concentrated than the accelerator layer, and for most of the period it was concentrated in vendors who also supplied the machine: Cray's SeaStar, Gemini and Aries; IBM's torus networks; Fujitsu's Tofu. The independent merchant fabric is a relatively recent and relatively brief phenomenon.

The second is that the 2019–2020 acquisitions removed the last two large independent fabric businesses in the same twelve months. After them, the two companies that supply most accelerator FLOPS also supply most fabric FLOPS. Accelerator share and fabric share stopped being independent variables, which matters to anyone modelling concentration risk from either one alone.

The third is Omni-Path, which is the counterexample worth keeping. Intel discontinued it in 2019 and it did not die: Cornelis Networks took it over in 2020 and continues it. It is the only case in this dataset of a fabric outliving its vendor's interest in it, and it is why the parts table records Cornelis as the current supplier of record while the edges from Stampede2, SuperMUC-NG and TSUBAME3.0 still correctly name Intel as the supplier at build.