TPU v6e — specs & cloud pricing
The TPU v6e is a datacenter accelerator from Google. Live specs and modeled cloud $/hr from the Mintok silicon catalog.
Cloud rate
$2.70per chip-hour (modeled)
Specifications
- Vendor
- Class
- Datacenter accelerator
- Architecture
- TPU
- BF16 throughput
- 918 TFLOPS
- On-package memory
- 32 GB
- Memory bandwidth
- 1,638 GB/s
- TDP
- 600 W
More Google silicon
FAQ
How much does the TPU v6e cost per hour in the cloud?
The TPU v6e rents for about $2.70 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 TPU v6e have?
The TPU v6e has 32 GB of on-package memory at 1,638 GB/s bandwidth.
What is the TPU v6e's BF16 compute throughput?
The TPU v6e delivers 918 BF16 TFLOPS at a 600 W TDP.
Is the TPU v6e a datacenter or edge accelerator?
The TPU v6e is a datacenter accelerator (600 W TDP).
Cost this chip in a fleet
Size a deployment on the TPU v6e, 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.