AI Chips & Semiconductor Hardware
This category covers the companies designing the silicon that AI actually runs on — GPUs, custom accelerators, and the interconnects that tie thousands of them together into a training cluster. Nvidia dominates the category by revenue, but the more interesting builder-relevant story is the challengers: Cerebras, Groq, SambaNova, and Tenstorrent are each betting on a different architecture to beat Nvidia on inference cost or latency for a specific workload. For most AI builders this category is invisible — you rent compute, you don’t buy chips — but the price and availability of that compute, and how fast new model architectures get supported, traces directly back to what’s happening here. GROUNDING tracks new chip announcements, benchmark claims, and supply dynamics that eventually show up as price changes one layer up.
At a glance
- 2 tracked companies (2 hand-curated)
- Most recently updated: AMD (2026-06-04)
Most actively covered
FAQ
What is the AI Chips & Semiconductor Hardware category?
Designs or fabricates silicon for AI training or inference.
How many companies are in this category?
GROUNDING currently tracks 2 companies in AI Chips & Semiconductor Hardware, 2 of which have a hand-curated profile.
How current is this hub?
The most recently updated entry is AMD, last mentioned 2026-06-04; the radar refreshes hourly.