Google unveils TPU 8tx and TPU 8i Max for AI workloads
Original title: Google broadens its AI hardware portfolio with two new specialized chips
TPU 8tx, designed with AMD, targets confidential training and inference; TPU 8i Max doubles memory in one package.
Google unveiled two eighth-generation tensor processing unit variants, TPU 8tx and TPU 8i Max, for its cloud AI hardware portfolio. TPU 8tx, designed with AMD and based on TPU 8t, is meant to keep data and model weights isolated during use and to run reasoning alongside training. TPU 8i Max pairs two April TPU 8i chips, doubling memory, processing power and bandwidth so models can handle more information per query.
Unconfirmed: A personal claim about Nvidia's CEO in the article is unrelated to the chip launch and is not independently confirmed here.
- TPU 8tx and TPU 8i Max extend Google's eighth-generation chips introduced in April.
- TPU 8tx includes a custom inferencing network for on-the-fly distillation and small-model training.
- Google says the launch reduces reliance on chips from other companies for cloud AI.
- The article also cites custom-chip work by Microsoft, Amazon, OpenAI and Anthropic.
- Products
- TPU 8txTPU 8i MaxTPU 8tTPU 8i
From the original article
2 min read at techcrunch.com
Google's new tensor processing units (TPUs) broaden its chip portfolio to more use cases: confidential model training and more memory capacity.
By Ram Iyer · techcrunch.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.