Supercomputers/Systems/KDDI GPU Cloud
Japan · operational · ai training
KDDI GPU Cloud
Also known as KDDI Osaka Sakai GB200 NVL72
Operated by KDDI at KDDI Osaka Sakai Data Center .
Phase 1 seed dataset, compiled by hand. These rows were built from public operator, laboratory and vendor sources. A mechanical second-reader pass has since fetched every cited source: 1092 of 1153 systems have a readable citation that names them, and 40 are genuinely weakly sourced. Every claim carries its source and a confidence tier. Treat anything below verified as a lead, not a citation.
Measured performance
- Rmax
- 22.21 PFlop/s
- Rpeak
- 24.12 PFlop/s
- Rmax ÷ Rpeak
- 92.1%
- Cores
- 148,896
- Power
- -
- Per watt
- - GF/W
Figures are as last publicly reported for the configuration described below, not a live measurement. Where a system was upgraded in place, the post-upgrade configuration is shown and the earlier one appears in the timeline.
What this machine is made of
One row per supplier relationship. “Supplier at build” is the company that shipped the part at the time; where that company has since been acquired, the parent it rolls up to today is shown beside it. That distinction is what makes ticker-level aggregation possible across a thirty-year dataset.
| Role | Part | Supplier at build | Quantity | Confidence | Source |
|---|---|---|---|---|---|
| Integrator | - NVIDIA DGX GB200 NVL72 (manufacturer listed as HPE) | Hewlett Packard Enterprise | - | Reported | top500.org |
| Accelerator | NVIDIA GB200 NVL72 Accelerators Arm CPU + CUDA GPU · TSMC 4NP NVIDIA GB200 NVL72 | NVIDIA | - | Verified | top500.orgnewsroom.kddi.com |
| CPU | NVIDIA Grace CPUs Armv9 (Neoverse V2) · 72C · 3.1 GHz · TSMC 4N NVIDIA Grace 72C 3.1GHz | NVIDIA | - | Reported | top500.org |
| Interconnect | - NVIDIA Spectrum-X Ethernet | NVIDIA | - | Reported | top500.org |
| Operating system | Ubuntu Server Operating systems Linux distribution · Open source, free with paid support tiers Ubuntu 24.04.1 LTS | Canonical | - | Reported | top500.org |
Compute deals
- 2025-06-25: KDDI and Hewlett Packard Enterprise (chip supply ). telecomtv.comnewsroom.kddi.com
All compute deals. Headline values are as announced, not additive.
The read
KDDI's GPU cloud service on NVIDIA GB200 NVL72 servers at its Osaka Sakai Data Center, the former Sharp Sakai plant KDDI acquired in April 2025. KDDI's newsroom release of 22 January 2026 states trial operations began that day, applications opened from 1 April 2026, and the four-storey, roughly 57,000 square metre building uses direct liquid cooling plus air cooling and renewable power; it does not give a GPU count. TOP500 lists the system as an NVIDIA DGX GB200 NVL72 (manufacturer HPE) with NVIDIA Grace 72C 3.1GHz CPUs, Spectrum-X Ethernet and Ubuntu 24.04.1, 148,896 cores, Rmax 22.21 PFlop/s and Rpeak 24.12 PFlop/s at rank 98 on the June 2026 list. The TOP500 June 2026 edition is full in this dataset, so no list appearance is recorded and ranked stays 0.
Timeline
- 2026-01 First operational source
Site & facility
KDDI Osaka Sakai Data Center
- Location
- Sakai, Osaka, Japan
- Cooling
- Direct liquid cooling plus air cooling
- Commissioned
- 2026
Change history
- 2026-10-04 Added KDDI GPU Cloud, GB200 NVL72 at the Osaka Sakai Data Center
Source check
We fetched this system's own citations and recorded whether each page actually mentions it. This is a corroboration signal, not a fact check, and it is published so you can see how well the sourcing holds up rather than take it on trust.
| Cited source | Result | Found on the page |
|---|---|---|
| newsroom.kddi.com | names system and part | NVIDIA GB200 NVL72 |
| top500.org | names system and part | NVIDIA GB200 NVL72, NVIDIA Grace, Ubuntu Server |
Checked 2026-10-09 by pnpm verify. Re-run it and the table changes with the web.
Further reading
Operating a supercomputer
Acceptance testing, why "installed" and "in production" are six months apart, and the machine lifecycle.
Inside a node
Why accelerators exist, what the memory hierarchy costs, and how the CPU and accelerator merged onto one package.
The interconnect
Topologies, why latency and tail behaviour matter more than bandwidth, and how the fabric market consolidated.