Google's TPU 8tx is the more interesting half of the November 9 announcement, because confidential inference gives enterprise buyers a…
Google's TPU 8tx is the more interesting half of the November 9 announcement, because confidential inference gives enterprise buyers a security reason to pick Google over Nvidia. What would change my mind on Nvidia is a named customer moving secure workloads onto 8tx.
- Google announced two additions to its eighth-generation TPU portfolio. Bloomberg and Reuters describe them as reconfigured versions of existing systems. 8tx uses the same silicon as TPU 8t with custom AMD CPU hosts.
- Google says 8tx is optimized for secure, confidential inference and small-scale training, with up to 2x lower latency than TPU 8t. TPU 8i Max pairs two 8i chips for large context windows, doubling compute, SRAM and HBM bandwidth versus TPU 8i, per Google.
- Google's post names no customer, pricing or availability date for either system. It does say its broader TPU business has nearly 100 customers.
- $GOOGL was at $291.90 premarket November 9, up 2.5% from the November 6 close of $284.78.
My lean on Nvidia stays mixed. It is woven into everything AI, and these are Google's own performance claims with no outside validation yet. I'd flip if enterprises start committing to 8tx on price, since cheaper secure inference is exactly the kind of shift that makes Nvidia's product too expensive for part of the market. I could be wrong on how fast buyers move.
Sources:
Reuters: https://www.reuters.com/business/google-to-introduce-two-variants-its-in-house-chips-bloomberg-news-reports-2026-11-09/
Investing.com: https://ca.investing.com/news/stock-market-news/google-expands-ai-chip-lineup-as-nvidia-faces-ceo-spy-scandal-4942992
Benzinga: https://www.benzinga.com/news/26/11/63028551/google-escalates-nvidia-rivalry-specialized-versions-ai-chips-bloomberg-exclusive
Not financial advice. For informational and educational purposes only.