AI Research #Google #TPU #AI chips #Nvidia

Google Cloud Launches TPU 8t and 8i AI Chips

New custom AI chips offer 3x faster training and 80% better performance per dollar, directly competing with Nvidia.

Wednesday April 22, 2026

TL;DR — Quick Answer

Google Cloud has announced its eighth generation of custom AI chips, split into TPU 8t for training and TPU 8i for inference. The new chips promise 3x faster AI model training and 80% better performance per dollar, positioning Google Cloud squarely against Nvidia in AI compute.

Key Takeaways

→ TPU 8t is designed for training and TPU 8i for inference
→ Up to 3x faster AI model training claimed by Google Cloud
→ 80% better performance per dollar versus previous chip versions
→ Supports clusters of 1 million+ TPUs
→ Google Cloud will offer both TPUs and Nvidia's Vera Rubin chips later in 2026

Google Cloud Launches TPU 8t and 8i AI Chips — AI news article illustration

Google Cloud has announced its eighth generation of custom AI chips, split into two models: TPU 8t for training and TPU 8i for inference.

Performance Claims

3x faster AI model training
80% better performance per dollar
1 million+ TPUs in a single cluster
Reduced energy and cost compared to previous versions

Competition with Nvidia

Google’s chips don’t replace Nvidia—they supplement it. Google Cloud will offer both TPUs and Nvidia’s upcoming Vera Rubin chip later in 2026.

Partnership

Google and Nvidia are collaborating on networking technology called Falcon to improve efficiency in Google Cloud.

Enterprise Impact

The chips offer enterprises more affordable AI compute, potentially lowering entry barriers for AI development.

Written by Massin BSN

Frequently Asked Questions

What are the two new TPU models?

TPU 8t for AI model training and TPU 8i for inference.

How do Google's chips compare with Nvidia?

Google says they supplement rather than replace Nvidia, and Google Cloud will offer both TPUs and Nvidia's Vera Rubin chip later in 2026.

What is the Falcon partnership?

Google and Nvidia are collaborating on Falcon, a networking technology to improve efficiency in Google Cloud.

AI Research