Definition
AI chip makers are the companies that design the specialized processors that power artificial-intelligence training and everyday AI services.
At a glance
- Nvidia is the runaway leader, with roughly 80-85% of data-center AI chips by revenue. [1]
- AMD is the clear number two and fastest-growing challenger, near 10-12%.
- Cloud giants like Google design their own chips (TPUs) to cut costs and depend less on Nvidia. [3]
- Almost all of these chips, whatever the brand, are physically built by one factory: TSMC in Taiwan.
The list
- Nvidia — Dominant supplier; its GPUs power most AI worldwide. [4] (~80-85% share; $115.2B data-center revenue FY2025)
- AMD — Main alternative to Nvidia and the fastest-growing rival. [1] (~10-12% share)
- Google — Designs its own TPU chips for its data centers. [3] (~4.3M units projected 2026)
- Broadcom — Quietly co-designs the custom chips behind Google and others.
- Intel — Veteran chip maker still building an AI foothold behind the leaders. [2]
How to read this
Two numbers matter: market share (how much of the business a company wins) and revenue (the actual dollars). AMD and Intel sell chips to everyone, like Nvidia. Google and Broadcom are different: Google builds chips for its own data centers, and Broadcom helps turn those designs into working silicon.
Bottom line
The market is lopsided — Nvidia supplies most data-center AI chips, AMD trails far behind, and the trend to watch is cloud firms building custom silicon to depend less on one supplier.