Inferentia2 — specs & cloud pricing
The Inferentia2 is a datacenter accelerator from AWS. Live specs and modeled cloud $/hr from the Mintok silicon catalog.
Cloud rate
$0.55per chip-hour (modeled)
Specifications
- Vendor
- AWS
- Class
- Datacenter accelerator
- Architecture
- Custom
- BF16 throughput
- 190 TFLOPS
- On-package memory
- 32 GB
- Memory bandwidth
- 820 GB/s
- TDP
- 200 W
More AWS silicon
FAQ
How much does the Inferentia2 cost per hour in the cloud?
The Inferentia2 rents for about $0.55 per chip-hour on cloud providers (modeled from public rates). Own-vs-rent economics — capex, lease and TCO — are in the Mintok sizing report and rack teardowns.
How much memory does the Inferentia2 have?
The Inferentia2 has 32 GB of on-package memory at 820 GB/s bandwidth.
What is the Inferentia2's BF16 compute throughput?
The Inferentia2 delivers 190 BF16 TFLOPS at a 200 W TDP.
Is the Inferentia2 a datacenter or edge accelerator?
The Inferentia2 is a datacenter accelerator (200 W TDP).
Cost this chip in a fleet
Size a deployment on the Inferentia2, or see the full priced rack BOM it sits in.
Specs and modeled rates maintained by Mintok's catalog watchers from public sources. Not investment advice. Methodology.