👋 Take the State of AI Report Survey!

COMPUTE INDEX SOURCES.

🏠 Compute Index State of AI

This page documents the source for each data point in the State of AI Report Compute Index charts. Data cutoff: 1 October 2026. GPU-count charts show NVIDIA data-center GPUs (GH200/GB200/GB300 counted as GPU dies; one NVL72 rack = 72 GPUs), split into Deployed / Installing / Announced. Figures are charted by owner/operator; renters are listed separately under the demand-side chart and are not double-counted. National-HPC counts are exact (published); large company, neocloud and sovereign figures are best-available estimates or announced commitments. Where a primary source did not disclose a GPU count, none is invented.

Cited chip / NVIDIA / startup usage and Research topic signals

  • All four citation charts — Zeta Alpha analysis of open-source AI papers (per-year citation counts by chip family, by NVIDIA chip, by AI-chip startup, and chip-topic skew vs corpus baseline). Figures for 2026 are counts as of 1 September 2026 plus H2 predictions adjusted for volume. The research-topic dataset covers 6,356 papers (Jan 1 to June 1, 2025), topic-labelled with GPT-4o-mini. (link)

A100 GPU count (Ampere)

  • Meta Research SuperCluster — 16,000 — Meta AI blog (link)
  • Leonardo Booster (CINECA) — 13,824 — CINECA / Top500 (link)
  • Perlmutter (NERSC) — 7,168 — NERSC architecture docs (link)
  • Polaris (Argonne) — 2,240 — ALCF (link)
  • SoftBank — ~2,000 Ampere (DGX SuperPOD, operating since Sep 2023; SoftBank says 'approximately 2,000 NVIDIA Ampere GPUs') — SoftBank (link)
  • Berzelius (Linkoping) — 752 — Prior State of AI Compute Index
  • Alps (CSCS) — 576 A100 (144 nodes x 4) — CSCS (link)
  • Delta (NCSA) — 440 — NCSA Delta docs (link)
  • Jean Zay (IDRIS) — 416 — IDRIS (link)
  • NSCC ASPIRE 2A (Singapore) — 352 — NSCC (link)
  • Gadi (NCI) — 16 — NCI (link)

Hopper GPU count (H100 / H200 / GH200)

  • SpaceXAI Colossus 1 — 200,000 (150k H100 + 50k H200) — Data Center Dynamics; SemiAnalysis (link)
  • Tesla Cortex — ~66,000 (H100-equivalent) — Tesla shareholder deck / 8-K (link)
  • Meta GenAI clusters — 49,152 (2x 24,576 H100) — Meta Engineering blog (link)
  • CoreWeave (one cluster) — 42,000 H200 — Epoch AI; The Next Platform (link)
  • Voltage Park — 24,000 H100 — Data Center Dynamics (link)
  • JUPITER Booster (Julich) — 23,536 GH200 — EuroHPC JU; Top500 (June 2026) (link)
  • Microsoft Eagle — 14,400 H100 — Top500 (link)
  • Alps (CSCS) — 10,752 GH200 — CSCS; Top500 (link)
  • ChronoScale (Applied Digital) — 6,144 H100 (deployed; third-party colos in CO/MN/UT, finance-leased, operator not owner) — ChronoScale 10-K (FY to 31 May 2026) (link)
  • AIST ABCI 3.0 — 6,128 H200 — NVIDIA / AIST (link)
  • Isambard-AI (Bristol) — 5,448 GH200 — UKRI / University of Bristol (BriCS) (link)
  • NVIDIA Eos — 4,608 H100 — NVIDIA (link)
  • MareNostrum 5 ACC (BSC) — 4,480 H100 — Barcelona Supercomputing Center (link)
  • KAUST Shaheen III — 2,800 GH200 — KAUST; Top500 (link)
  • Venado (LANL) — 2,560 GH200 — Top500 (link)
  • Israel-1 (NVIDIA) — 2,048 H100 — NVIDIA (link)
  • NSCC ASPIRE 2B (Singapore) — 1,536 H200 — NSCC (link)
  • Gefion (DCAI, Denmark) — 1,528 H100 — NVIDIA; Novo Nordisk Foundation (DCAI is owner: NNF 85% / EIFO 15%) (link)
  • Arrhenius (NAISS, Sweden) — 1,528 GH200 (deployed; entered production early June 2026, Top500 #42) — EuroHPC JU supercomputer register (publishes the 1,528 total); NAISS (production date) (link)
  • Jean Zay (IDRIS) — 1,456 H100 — IDRIS (link)
  • Leonardo LISA upgrade (CINECA) — 1,328 H100, 166 8-way servers (installing; inaugurated 11 Jun 2026, user availability expected summer 2026) — EuroHPC JU (link)
  • Yotta Shakti Cloud (India) — 8,192 H100 (deployed; Nidar Infrastructure 424B3, Nov 2025: 'currently all 8,192 H100s are operational'. A further 2,048 H200 and 8,192 B200 were on order but the registration was withdrawn in Jan 2026 and neither is confirmed installed) — Nidar Infrastructure SEC 424B3 (link)
  • NSCC ASPIRE 2A+ (Singapore) — 320 H100 (40 DGX H100 nodes) — NSCC (link)
  • SoftBank — ~4,000 Hopper (deployed since Oct 2024; SoftBank says 'approximately 4,000 NVIDIA Hopper GPUs') — SoftBank (link)
  • NCHC Nano4 (Taiwan) — 1,760 H200 (220 nodes x 8; in formal operation from 1 Jul 2026) — NCHC; CNA (link)
  • DeltaAI (NCSA) — 608 GH200 — NCSA Delta documentation (link)

Blackwell GPU count (B200 / B300)

  • Deutsche Telekom Industrial AI Cloud (Munich) — up to 10,000 B200 — Deutsche Telekom (link)
  • SoftBank DGX SuperPOD — >4,000 B200 — SoftBank (link)
  • Naver Cloud — 4,000 B200 (deployed) — Data Center Dynamics (link)
  • E2E Networks (India) — 1,024 B200 — E2E Networks (link)
  • SK Telecom Haein — >1,000 B200 — Data Center Dynamics (link)
  • IREN — 50,000+ B300 (announced; purchase agreements, phased delivery through H2 2026) — IREN (link)
  • Yotta D2 (Greater Noida, India) — 20,736 HGX B300 (announced; go-live targeted Aug 2026) — Yotta (link)
  • Roche — 2,176 Blackwell on premises across the US and Europe (deployed; the separate cloud tranche taking Roche 'over 3,500' is excluded) — Roche media release, 16 Mar 2026 (link)
  • WhiteFiber (Bit Digital) — 2,096 B300 (announced: 1,392 BaseTen Nov 2026 + 576 Iceland Dec 2026 + 128 ordered Jul 2026) — WhiteFiber 10-Q (link)
  • Eli Lilly (LillyPod, Indianapolis) — 1,016 B300 (DGX SuperPOD, deployed) — NVIDIA (link)
  • HIVE / BUZZ HPC (Bell AI Fabric, Manitoba) — 504 B200 (deployed; went live May 2026) — HIVE Digital FY2026 results (link)
  • NxtGen (India) — over 4,000 Blackwell in Dell PowerEdge XE9685L servers (announced; B200 vs B300 unstated, no operating confirmation) — CIOL quoting Dell, 28 Jan 2026; Vertiv release, 18 Feb 2026 (link)
  • WhiteFiber (Bit Digital) — 384 B200 (deployed; service order with 'Deployment and revenue generation commenced in January 2026'. A minimum: total B200 fleet undisclosed) — WhiteFiber 10-Q (link)
  • NHN Cloud (Korea) — 7,656 B200 (deployed; two NIPA clusters of 6,120 procured under the MSIT program and operated by NHN, plus ~1,536 for NHN's own use. Part of the South Korea aggregate on the Grace-Blackwell chart) — Asia Economy, 7 Aug 2026; Top500 (link)
  • Firebird DC-1 (Armenia) — 6,144 B200 (deployed; operational debut 8 Aug 2026) — Firebird (link)
  • Datasection — 5,080 B300 installing (Inzai, Japan) + 10,000 B300 announced (Sydney) + 4,696 B200 announced (Bangkok) — Datasection releases (link)
  • Nebius Israel (Modiin) — 4,000 B200 (deployed; Nebius: 'In October we launched in Israel with a deployment of NVIDIA B200s'; the 4,000 is from the launch release) — Nebius (link)
  • Kakao (Korea) — 2,040 B200 (deployed; government-procured, Kakao-operated, 84% of its 2,424 allocation) — Asia Economy (link)
  • Cloud GA instances (Google A4, AWS P6, Azure, Lambda, RunPod, DataCrunch) — Vendor product pages; counts not disclosed

Grace-Blackwell GPU count (GB200 / GB300)

  • Stargate Abilene (Oracle/Crusoe) — at least 131,000 installing / 450,000 target (installing: Oracle 'delivered 131 thousand GPUs there in Q1' FY2027, Jun-Aug 2026, and 6 of 8 buildings 'delivered to the customer'. A one-quarter delivery figure, so a floor, chip not named on the call; delivery is not counted as operation. Oracle owns the silicon and leases to OpenAI; Crusoe is builder/landlord) — Oracle Q1 FY2027 earnings call, 10 Sep 2026; Data Center Dynamics (450,000 target) (link)
  • HUMAIN (Saudi) — 18,000 GB300 phase 1 / up to 600,000 — NVIDIA newsroom; Middle East AI News (link)
  • South Korea (national + industry) — 260,000 (announced; a country-wide aggregate. The sovereign slice is MSIT 'up to 50,000' from an 'initial deployment of 13,000'; the rest is Samsung, SK Group, Hyundai and NAVER Cloud, and the mix includes RTX PRO 6000 Blackwell Server Edition as well as GB-series) — NVIDIA newsroom (link)
  • Nscale to Microsoft — 200,000 GB300 (announced) — Nscale (link)
  • SpaceXAI Colossus 2 — 550,000 deployed (110,000 GB200 + 440,000 GB300), 440,000 GB300 announced (220,000 targeted for the week of September 28 and 220,000 for November). The post does not establish that the first future tranche is racked or energizing, and its operational target is not treated as completion. Sourced to a CEO post, not a filing: 'Colossus 2 is 110k GB200 and 440k GB300.' The June prospectus (424B4 p.76) had 220,000 brought online. A further conditional December tranche is excluded. Google, Anthropic and Reflection are renters and carry no bar — Elon Musk on X, 25 Sep 2026; SpaceX 424B4 (12 Jun 2026) (link)
  • Argonne Solstice + Equinox (DOE) — 100,000 + 10,000 Blackwell (both announced; Argonne: 'Both systems are in the planning stages'. Announced as Blackwell, not explicitly Grace-Blackwell) — Argonne; NVIDIA newsroom (counts) (link)
  • Together / Hypertec — 36,000 GB200 — Together AI (link)
  • Not charted: Aolani Cloud (Malaysia, ByteDance as renter) — the WSJ-reported 36,000 GB200 plan was retired from the bars in October 2026. Aolani's own release (21 Sep 2026) says 22,000 new GPUs in early 2027 'will bring Aolani's total deployment to over 48,000 GPUs', which leaves no room for a 36,000 Blackwell cluster today — Aolani press release, 21 Sep 2026 (link)
  • Stargate UAE / G42 — 35,000 GB300 (installing) — OpenAI (link)
  • SpaceXAI Colossus 1 — 30,000 GB200 — SemiAnalysis (link)
  • AWS Project Ceiba — 20,736 GB200 (288 NVL72 racks) — AWS; Data Center Dynamics (link)
  • UK Nebius/Nscale — 14,000 Blackwell (announced) — NVIDIA newsroom (link)
  • CoreWeave — 8,192 GB300 NVL72 (deployed; MLPerf Training v6.0) — CoreWeave (link)
  • Mistral (Eclairion, Bruyeres-le-Chatel, France) — 13,800 GB300 (installing; CTO in Mar 2026: 'on stabilise actuellement la premiere tranche'. No operating confirmation after the end-June 2026 target) — silicon.fr; Data Center Dynamics (link)
  • Sines / Start Campus (Nscale/Microsoft, PT) — over 12,600 GB300 (deployed; Nscale's S-1 lists the Portugal site as Active, defined as online and revenue generating. Microsoft is the offtaker and carries no bar) — Nscale S-1, 18 Sep 2026 (link)
  • Lambda — 10,000 GB300 NVL72 (deployed; Lambda: 'in production, on a 10,000-GPU NVIDIA GB300 NVL72 cluster'. Lambda owns and operates it; Microsoft and other customers are renters) — Lambda blog, 31 Mar 2026 (link)
  • Nscale Glomfjord (Norway) — 2,304 GB200 (32 NVL72; announced: Nscale owns the GPUs, service was due to start 30 Aug 2025 with no confirmation it did. Renter is Spring (SG) Pte. Ltd.) — Nscale S-1, exhibit 10.12 (link)
  • MareNostrum 5 AI Factory upgrade (BSC) — 1,920 (336 nodes x 4 GB200 NVL4 + 144 nodes x 4 GB300; announced, H1 2026 install target lapsed unconfirmed) — Barcelona Supercomputing Center (link)
  • Sferical AI (Linkoping) — 1,152 GB300 (16 GB300 NVL72 racks; announced, owned by Sferical AB, no operating confirmation) — Sferical AI (link)
  • NCHC Nano4 GB200 (Taiwan) — 144 (2 GB200 NVL72; deployed, opened to users as a free trial on 1 Sep 2026) — NCHC (link)
  • Taiwan Foxconn 'Big Innovation' — 10,000 GB300 (announced) — NVIDIA newsroom (link)
  • IT4LIA AI Factory (Bologna) — 8,000+ GB200 NVL4 (announced) — EuroHPC JU (link)
  • Microsoft Azure NDv6 GB300 — 4,608 (for OpenAI) — NVIDIA newsroom (link)
  • Jane Street — 4,032 GB300 (56x GB300 NVL72, owned Texas facility; deployed) — NVIDIA AI Infrastructure (link)
  • Microsoft Azure NDv6 GB200 — ~4,000 — Microsoft Azure blog (link)
  • Nscale Keflavik (Verne campus, Iceland) — 4,608 GB300 (deployed; 64 GB300 NVL72 racks, benchmark suite run across all 64) — Nscale, 16 Jul 2026 (link)
  • HIVE / BUZZ HPC (Merritt BC, Bell/Cohere) — 2,304 GB200 (announced; contracted not live, 'go live in late 2026 to early 2027') — HIVE Digital Q1 FY2027 (link)
  • HIVE / BUZZ HPC (Merritt BC, GB300) — 2,016 GB300 NVL72 (announced; separate cluster and customer, delivery calendar Q4 2026, HIVE retains ownership) — HIVE Digital 8-K, 19 Aug 2026 (link)
  • Blue Swan / BavariaAI (FAU Erlangen + LRZ) — 1,024 GPUs via GB200 NVL4 (announced; FAU: 'a further 1,024 AI chips from... NVIDIA', contract signed 23 Apr 2026, access targeted autumn 2026. NVIDIA rounded to 1,000) — FAU; NVIDIA newsroom (ISC, 22 Jun 2026) (link)
  • HammerHAI (HLRS Stuttgart) — 860 GPUs (215 GB200 NVL4 nodes; announced, in service H2 2026) — HLRS (link)
  • Mimer (EuroHPC JU, hosted at Linkoping) — 400 GPUs (100 GB200 NVL4 systems; announced) — NVIDIA newsroom (ISC, 22 Jun 2026) (link)
  • NERSC Cech (LBNL) — 144 Blackwell with 72 Grace CPUs (installing; early-access system for Doudna) — NERSC (link)
  • WhiteFiber (Bit Digital) — 72 GB200 (deployed; revenue generation began 7 Mar 2026. A further 216 GB200 remain unconfirmed) — WhiteFiber 10-Q (link)
  • Indosat (Indonesia) — 2,304 GB200 — Tom's Hardware (link)
  • SoftBank — 1,224 GB200 — SoftBank (link)

Frontier-lab contracted compute (GW) — demand side

  • OpenAI — NVIDIA 10 GW / up to $100B — NVIDIA newsroom (link)
  • OpenAI — Oracle $300B (~4.5 GW, Stargate) — OpenAI; CNBC (link)
  • OpenAI — Microsoft Azure $250B; AWS $38B; CoreWeave $22.4B — CNBC ($1T AI deals guide) (link)
  • OpenAI — AMD 6 GW / MI450 (non-NVIDIA) — OpenAI; TechCrunch (link)
  • OpenAI — Broadcom 10 GW of OpenAI-designed accelerators across multiple generations to 2029 (non-NVIDIA; term sheet, Oct 2025). 'Jalapeno' is the first of those generations and carries no GW figure of its own — OpenAI / Broadcom (link)
  • OpenAI — AWS ~2 GW Trainium3/Trainium4 (non-NVIDIA; 27 Feb 2026, inside a $100B/8yr expansion of the $38B AWS agreement) — Amazon press release (link)
  • OpenAI — Cerebras 750 MW wafer-scale (non-NVIDIA; signed 14 Jan 2026) — Cerebras (link)
  • Anthropic — Google + Broadcom 5 GW of next-generation TPU (non-NVIDIA; Anthropic's own May 2026 restatement. Broadcom supplies ~3.5 GW of that total). Builds on the Oct 2025 'well over a gigawatt' Google Cloud TPU expansion, which neither party restates as a separate addend — Anthropic (link)
  • Anthropic — AWS Project Rainier ~500k-1M Trainium2 (non-NVIDIA) — Anthropic (link)
  • Anthropic — Microsoft Azure + NVIDIA: $30B of Azure compute plus additional capacity up to 1 GW (the 1 GW is additional to the $30B, not a GW-denomination of it) — Anthropic (link)
  • Anthropic — AMD up to 2 GW / MI450 (non-NVIDIA; 22 Jul 2026, AMD equity up to $5B, deploys from H1 2027) — AMD investor relations (link)
  • Anthropic — SpaceXAI (Colossus 1) lease: 'more than 300 megawatts of new capacity (over 220,000 NVIDIA GPUs)'. The $1.25B/mo price is press-reported, not in either party's announcement — Anthropic (link)
  • Microsoft (offtaker) — Nscale 200k GB300; IREN 200 MW/$9.7B; Lambda; Nebius; Crusoe 900 MW — Operator press releases
  • Meta (offtaker) — CoreWeave $35.2B; Nebius up to $27B — The Globe and Mail; press (link)
  • Startups (Cohere/Runway->CoreWeave; Thinking Machines/World Labs->Google Cloud; Reka/Suno->Oracle; Sarvam->Yotta 4,096 H100; Upstage->SKT ~500 B200; Inflection->CoreWeave 22k legacy) — Respective cloud case studies / press

Compiled October 2026. Citation-chart data: Zeta Alpha (last refreshed 1 September 2026). Compute-cluster data: operator disclosures, EuroHPC, Top500 (June 2026), SemiAnalysis, The Next Platform, Data Center Dynamics, company earnings and SEC filings, and NVIDIA newsroom. Estimates and announced commitments are labelled as such in the underlying dataset.