Updated May 2, 2026

AI Power Rankings

Independent·Frontier AI lab index
Reference · Updated May 1, 2026

AI Chipmakers

The companies that build the silicon AI runs on. They are critical infrastructure for the entire ecosystem, but they are evaluated separately from the AI labs ranked in the main index — different game, different rubric.

Note: some labs in the main ranking (Google, Amazon, Meta) also design their own AI silicon — TPUs, Trainium, MTIA — but their primary role is building models and products, so they're scored there, not here.

1. NVIDIA

  • HQ: Santa Clara, CA
  • Key chips: H100, B200, Blackwell, Grace Hopper
  • Key software: CUDA, TensorRT, NIM microservices
  • Notes: Dominant AI GPU provider; CUDA ecosystem is a massive moat; essential to virtually every AI lab's training stack

2. AMD

  • HQ: Santa Clara, CA
  • Key chips: MI300X, MI350, Instinct series
  • Notes: Primary challenger to NVIDIA in AI GPUs; gaining traction in data center AI workloads

3. Intel

  • HQ: Santa Clara, CA
  • Key chips: Gaudi 3, Xeon with AI acceleration
  • Notes: Pushing AI accelerators and foundry services; large enterprise footprint but trailing in AI GPU adoption

4. Broadcom

  • HQ: San Jose, CA
  • Key chips: Custom AI ASICs, networking silicon (Jericho, Memory, Fabric)
  • Notes: Designs custom AI chips for hyperscalers (Google TPUs, Meta MTIA); dominant in AI networking interconnects

5. Qualcomm

  • HQ: San Diego, CA
  • Key chips: Snapdragon X Elite, Cloud AI 100
  • Notes: Leading edge AI on mobile and PC devices; pushing into on-device inference

6. Cerebras

  • HQ: Sunnyvale, CA
  • Key chips: WSE-3 (Wafer Scale Engine)
  • Notes: Novel wafer-scale approach; targeting training and inference at massive scale

7. Groq

  • HQ: Mountain View, CA
  • Key chips: LPU (Language Processing Unit)
  • Notes: Inference-focused; known for extremely fast token generation speeds

8. SambaNova

  • HQ: Palo Alto, CA
  • Key chips: SN40L, DataScale systems
  • Notes: Enterprise AI hardware+software platform; reconfigurable dataflow architecture

9. TSMC

  • HQ: Hsinchu, Taiwan
  • Key role: Fabrication of virtually all leading AI chips (NVIDIA, AMD, Apple, Qualcomm, Cerebras, etc.)
  • Notes: Not a chip designer but the essential manufacturing backbone; advanced process nodes (3nm, 2nm) are a bottleneck for the entire industry

This list covers the key companies competing in AI silicon — from GPU giants to specialized accelerator startups to the foundry that makes it all possible.