 Vera RubinRubin GPU, Vera CPU, NVL72 rack, HBM4 |  NVIDIA | Deploy | Partner cloud shipments | Q4 2026 | 16d ago |
 TeraPHY optical I/OOptical chiplet, UCIe |  Ayar Labs | Pilot | Volume production | — | 30d ago |
 Starcloud satellitesStarcloud-1 H100, Starcloud-2 Blackwell |  Starcloud | Proto | Starcloud-2 launch | — | 2mo ago |
 CoreWeave CloudGPU clusters, GB200/GB300 racks |  CoreWeave | Scaling | More than 1.85 GW active power | Dec 2026 | 2mo ago |
 Low-NA EUV (NXE)Standard EUV, NXE:3800E |  ASML | Scaling | | — | 3mo ago |
 Prometheus and HyperionGigawatt AI data centers |  Meta | Deploy | Prometheus online | 2026 | 3mo ago |
 Instinct MI450 / HeliosMI455X GPU, 72-GPU rack, HBM4 |  AMD | Pilot | First OpenAI Helios capacity online | Q4 2026 | 3mo ago |
 BlackholeAI processor, RISC-V |  Tenstorrent | Deploy | | — | 3mo ago |
 CoWoS advanced packagingChip packaging, HBM integration |  TSMC | Scaling | | — | 4mo ago |
 ColossusMemphis AI supercomputer |  xAI | Deploy | Colossus 2 gigawatt-scale build | — | 4mo ago |
| PassageM1000 optical base, L200 co-packaged optics |  Lightmatter | Pilot | Volume customer deployments | — | 4mo ago |
| Mistral ComputeEuropean AI cloud |  Mistral AI | Pilot | Initial data center capacity in France | — | 4mo ago |
 N2 process2 nm class, gate-all-around | TSMC | Scaling | N2P and A16 volume production | Q4 2026 | 6mo ago |
 Tensor Processing Unit (TPU)Google's AI chips, Ironwood |  Google DeepMind | Scaling | TPU 8t and TPU 8i generally available | 2026 | 6mo ago |
| HBM4Stacked AI memory, 2,048-bit interface |  SK hynix | Deploy | HBM4 volume supply for NVIDIA Rubin | 2026 | 7mo ago |
 WSE-3 / CS-3Wafer-scale chip, 900,000 cores |  Cerebras | Scaling | | — | 9mo ago |
 Intel 18AGate-all-around, backside power |  Intel | Scaling | Clearwater Forest Xeon 6+ ramp | 2026 | 9mo ago |
 LPU and GroqCloudFast inference chip, API cloud |  Groq | Deploy | | — | 10mo ago |
 TSMC ArizonaUS chip factories, Phoenix | TSMC | Deploy | Fab 2 production (N3) | 2027 | 10mo ago |
 High-NA EUV (EXE)High-NA EUV, EXE:5000, EXE:5200 | ASML | Pilot | High-volume manufacturing insertion (Intel 14A) | 2027 | 10mo ago |
 StargateAI data centers, 10 GW, Texas |  OpenAI | Deploy | Full ~10 GW US buildout | 2029 | 11mo ago |
 BlackwellB200, GB200 NVL72, GB300 | NVIDIA | Scaling | | — | 1y ago |
 Abilene AI campus1.2 GW Stargate site, Texas |  Crusoe | Deploy | Full campus complete | 2026 | 1y ago |
| Intel 14AHigh-NA EUV node | Intel | Research | Risk production | 2027 | 1y ago |
 Instinct MI300 and MI350MI300X, MI325X, MI355X GPUs | AMD | Scaling | | — | 1y ago |