Venado: every supplier in the machine

Los Alamos National Laboratory, United States

Accelerator: NVIDIA GH200 Grace Hopper Superchip (Hopper H100 GPU paired with a Grace CPU over NVLink-C2C), 2,560 in total
Host CPU: NVIDIA Grace: 72 Arm Neoverse V2 cores in each GH200; the CPU partition uses 920 Grace CPU Superchips of 144 Arm cores each (two Grace CPUs joined by NVLink-C2C)
Integrator: HPE (HPE Cray EX; TOP500 lists the system as HPE Cray EX254n)
Nodes: Node and cabinet counts not published; compute is 2,560 GH200 Superchips plus 920 Grace CPU Superchips
Memory: Grace CPU Superchip: up to 960 GB LPDDR5X at up to 1 TB/s. Capacity fitted to the GH200 modules not published
Interconnect: HPE Slingshot 11
Storage: Lustre parallel file system on the Slingshot network, provided under the HPE contract (reported); capacity and hardware not published
Cooling: Direct liquid cooling of the Grace Hopper Superchips in HPE Cray EX cabinets (the system is described as all liquid-cooled); facility plant not published
Power: 1.66 MW (1,661.57 kW, power measurement level 1, TOP500 June 2026 list)
Rmax (HPL): 98.51 PFlop/s (TOP500 June 2026 list)
Rpeak: 130.44 PFlop/s (TOP500 June 2026 list)
Status: Delivered to Los Alamos in March 2024 and opened at a ribbon-cutting on 15 April 2024 at the Nicholas C. Metropolis Center; moved to a classified network in 2025 and serves as a shared resource for the NNSA laboratories

Unmarked companies come from the operator's or a vendor's own pages (lanl.gov, energy.gov/nnsa, nvidia.com, hpe.com, an HPE-sponsored HPCwire article, arxiv.org papers co-authored by HPE engineers, top500.org); reported companies come from named trade press or analyst reporting whose article body was read in full, not just the headline. This is a map of who plays where, not a confirmed bill of materials: Los Alamos National Laboratory's April 2024 Venado announcement, the NNSA/DOE August 2025 article on Venado, NVIDIA's 2022 Grace announcement and architecture write-ups, an HPE-sponsored HPCwire article, HPE and arXiv Slingshot documentation, The Next Platform's 2022 and 2024 coverage, and the TOP500 entry (June 2026 list), researched 2026-10-10

This page is the parts-first view. For Venado's full system record on the sister site Compute Atlas, an open dataset of supercomputer and AI-cluster supply chains, see Venado on Compute Atlas.

System design and cabinets (HPE Cray EX, direct liquid cooled; designed and manufactured at HPE's Chippewa Falls, Wisconsin, facility and delivered to Los Alamos in March 2024)
Rack-scale systems 4.01
HPE (system integrator and prime contractor (HPE Cray EX); LANL describes it as an exascale-class HPE Cray EX system built through a codesign process with HPE and NVIDIA) · cabinet model and cabinet count (not published), makers of the cold plates, quick disconnects, coolant distribution units and rectifiers (not named in any source read)
Sources: lanl.gov · nextplatform.com · top500.org
Grace Hopper compute blades and nodes (2,560 GH200 Superchips, direct liquid cooled; TOP500 names the blade HPE Cray EX254n)
GPU baseboards & compute trays 4.03
HPE (HPE Cray EX254n is the blade name TOP500 gives for the system; the blade is not described on LANL's pages), Nvidia (GH200 Grace Hopper Superchips, 2,560 in the system) · superchips per node, nodes per blade and blades per cabinet (not published)
Sources: lanl.gov · top500.org · energy.gov
GH200 Grace Hopper Superchip (Arm-based Grace CPU plus Hopper H100 GPU joined by NVLink-C2C; 2,560 superchips)
Merchant GPUs / accelerators 1.01 · Server CPUs 1.05
Nvidia, Arm (Neoverse V2 core design used in the Grace CPU) · which companies made the HBM3 stacks and the LPDDR5X on the GH200 modules, and the HBM3 and LPDDR5X capacity fitted at Los Alamos (not named or stated in any source read), advanced-packaging flow and substrate suppliers for the GH200 module (none named in any source read)
Sources: lanl.gov · hpcwire.com · developer.nvidia.com
Grace CPU Superchip partition (920 CPU-only superchips, each two 72-core Grace CPUs, 144 Arm cores, linked by NVLink-C2C; up to 960 GB LPDDR5X per superchip)
Server CPUs 1.05
Nvidia (Grace CPU Superchip, the first large-scale deployment of the part in the United States per LANL), Arm (Neoverse V2 core design in the Grace CPU, per NVIDIA's Grace CPU Superchip page) · blade or node form factor for the Grace CPU Superchip partition (LANL, NNSA and HPE name only HPE Cray EX at system level), foundry for the Grace CPU die (not named in any source read), maker of the LPDDR5X DRAM (not named in any source read)
Sources: lanl.gov · hpcwire.com · nvidia.com
Hopper GPU die fabrication (TSMC 4N process)
Leading-edge foundry 1.09
TSMC (fabricates the GH100 die on its 4N process per NVIDIA's Hopper write-up; LANL's own pages do not name the foundry) · foundry for the Grace CPU die (not named in any source read)
Sources: developer.nvidia.com · lanl.gov
Network interface cards (HPE Cassini NIC in the Slingshot 11 interconnect)
NICs / SuperNICs 1.06
HPE (Cassini NIC is the Slingshot-11 adapter in HPE's documentation; LANL names only "HPE Slingshot 11 interconnect" for Venado) · NICs per node (not published)
Sources: lanl.gov · cpe.ext.hpe.com · top500.org
Slingshot 11 fabric (64-port Rosetta switch ASIC on TSMC 16 nm; 200 Gb/s per port per trade press)
Scale-out switch ASICs 3.02 · Switch systems & white-box ODMs 3.03 · Leading-edge foundry 1.09
HPE (designs and supplies the Slingshot switches; Venado is an HPE Cray EX system networked with Slingshot 11), TSMC (Rosetta switch ASIC process, per a paper co-authored by HPE engineers (the paper describes Slingshot generally, not Venado's switch count)) · switch counts, topology, and the cable and optics vendors for Venado's Slingshot links (not published)
Sources: lanl.gov · arxiv.org · nextplatform.com
Lustre parallel file system on the Slingshot network (provided under the HPE contract; product and capacity not published)
AI storage systems 2.10
HPE* (per The Next Platform, the Lustre storage cluster is part of HPE's contract; the article does not name the product (ClusterStor or otherwise) or the capacity) · storage hardware vendor, drive vendors and capacity (not stated in any source read), whether DeltaFS or other file systems were deployed alongside Lustre (the same article says LANL "is looking to also try out" DeltaFS "and perhaps others")
Sources: nextplatform.com · nextplatform.com (2)

Los Alamos, NNSA, HPE and NVIDIA name HPE and NVIDIA as the suppliers but publish almost nothing below that level. No source read gives Venado's cabinet count, node count, blade model for the Grace CPU Superchip partition, Slingshot topology or switch count, the HBM3 and LPDDR5X capacity actually fitted to the GH200 modules, the size or product name of the Lustre storage system, or any facility figure (cooling plant, water temperature, electrical feed or heat reuse) for the Nicholas C. Metropolis Center. No source read names the maker of the HBM3 stacks, the LPDDR5X DRAM, the cabinet cold plates, quick disconnects, coolant distribution units, rectifiers, the storage drives, or the Slingshot cables and optics, and none names the foundry for the Grace CPU die, so those parts are left off the lines. The TOP500 entry does not say which partitions the Linpack run covered. On funding and programme, The Next Platform's 2022 and 2024 articles report Venado as an experimental system paid for from the laboratory's own budget rather than a stockpile-programme workhorse, and the 2025 NNSA article calls it a shared resource for the NNSA laboratories; no source read labels it an Advanced Simulation and Computing (ASC) system. The "10 exaflops" figure in the 2022 announcements is an AI-precision peak (FP8 with sparsity, per NVIDIA as relayed by The Next Platform), not a Linpack figure, and the operator has not repeated it. LANL's own pages give conflicting installation years (see status_note). Power and cooling outside the cabinet sit in Layer 6 and Layer 7.

Machine-readable: Markdown ·JSON ·every line of every machine (CSV)

Other machines:El Capitan · Frontier · Aurora · JUPITER · Alps · LUMI · Leonardo · Isambard-AI · Perlmutter · Eagle (Microsoft Azure) · Eos · Fugaku · MareNostrum 5 · Horizon · Doudna (NERSC-10). See also the AI rack teardowns and thesupercomputer overview.