Google Launches TPU 8t, 8i Chips Targeting Nvidia’s $193.7B Data Center Market
Google debuted TPU 8t and TPU 8i AI processors optimized for model training and inferencing, offering 2.8x better price-to-performance than predecessors. Hyperscalers expanding in-house TPU capacity could pressure Nvidia’s data center segment, which accounted for $193.7 billion of its $215.9 billion fiscal 2026 sales.
1. Google Unveils TPU 8t and 8i Chips
Google introduced the TPU 8t, optimized for AI model training, and the TPU 8i, designed for inference and AI agents, with both processors slated for availability later this year.
2. Enhanced Price-to-Performance and Training Efficiency
The TPU 8t delivers a 2.8x improvement in price-to-performance over its predecessor and aims to reduce frontier model development cycles from months to weeks, addressing cost and time constraints for large-scale AI projects.
3. Competitive Pressure on Nvidia’s Data Center Business
Major cloud providers and AI labs are expanding in-house chip capacity, potentially encroaching on Nvidia’s data center segment, which generated $193.7 billion in fiscal 2026 and represents over 50% of its total sales.