Map / AI chips and networking
Nvidia NVDA
Dominant AI accelerator maker; Vera Rubin in full production; Q2 FY27 revenue $96.2B.
Depends on
- TSMC: Blackwell and Rubin chips. Analysts estimate Nvidia passed Apple as TSMC's largest customer in 2026, about 22% of revenue. MacRumors, 2026-01-28
- TSMC: CoWoS packaging capacity. Nvidia reportedly holds more than half of TSMC's 2026 CoWoS capacity. TrendForce, 2026-08-05
- Amkor: US packaging capacity. About $1.5B multi-year prepayment to expand Amkor's Peoria, Arizona site. ExplainX, 2026-07-27
- SK hynix: HBM4 for Vera Rubin. Roughly 60 to 70% of Nvidia's HBM4 volume. Tech Times, 2026-06-02
- Samsung Memory: HBM4 for Vera Rubin. An estimated 25 to 30% of volume. Tech Times, 2026-06-02
- Micron: HBM4 for Vera Rubin. Micron says HBM4 is shipping in volume for Vera Rubin. Tech Times, 2026-10-01
- Lumentum: Lasers for optical links. Nvidia paid about $2B each to Lumentum and Coherent for priority capacity. Tech Times, 2026-05-27
- Coherent: Lasers for optical links. Nvidia paid about $2B each to Lumentum and Coherent for priority capacity. Tech Times, 2026-05-27
- US Commerce (BIS): China export limits. H200 allowed case by case from January; Blackwell and Rubin stay banned. Interesting Engineering, 2026-01-13
- China regulators: H200 import block. Customs stopped H200 shipments despite US approval. Taipei Times, 2026-01-15
Supplies
- Intel: $5B equity stake. Nvidia bought $5B of Intel stock; Intel will design x86 CPUs linked to NVLink. Network World, 2025-09
- CoreWeave: Vera Rubin systems. Named as a Vera Rubin ramping partner. Nvidia, 2026-08-26
- Microsoft: Vera Rubin systems. Azure named as a Vera Rubin ramping partner. Nvidia, 2026-08-26
- Oracle: Vera Rubin systems. Oracle named as a Vera Rubin ramping partner. Nvidia, 2026-08-26
- Google (Alphabet): Vera Rubin systems. Google Cloud named as a Vera Rubin ramping partner. Nvidia, 2026-08-26
- Nebius: Vera Rubin systems. Nebius named as a Vera Rubin ramping partner. Nvidia, 2026-08-26
- SpaceXAI (xAI): GPUs for Colossus. About 555,000 GPUs across Colossus 1 and 2; 2 GW planned. Introl, 2026-01
- OpenAI: $30B equity plus dedicated capacity. Part of OpenAI's $110B February round; 3 GW inference plus 2 GW training on Vera Rubin. TechCrunch, 2026-02-27
Data as of 2026-10-01. Cite as: Ye, Mike. "Nvidia" (nvidia). AI Lattice, dataset v0.1.0. https://ailattice.ai/e/nvidia/. How to cite.