Supply chain / In-memory and wafer-scale
Cerebras
Builds AI processors that use an entire silicon wafer as one chip, with large on-chip memory, and sells access to them as a cloud service, a dedicated private service or on-premises systems.
- Type
- Company
- Where it sits
- A non-GPU architecture available now for fast model inference, useful where GPU supply or latency is a problem.
- Constraint
- Classed here under in-memory only because of its large on-chip memory; the register has no separate wafer-scale category.
- Maturity
- In use or open for access
What we checked
- The company page lists three ways to use its systems: public cloud endpoints with an API key, dedicated private cloud, and on-premises deployment. (source)
- Its chip page describes the current wafer-scale engine as 46,225 square millimetres with 900,000 cores and a design that tolerates manufacturing defects by routing around them. (source)
Written by ComputeFinder from public pages on 2 October 2026. Company statements are the company's own, not independently checked, and reported items are marked as such. Not yet reviewed line by line by a second person. If something is wrong, tell us.