Skip to content
NewA calmer, faster ZeroHour
StoryAI industryinfo

Google adds confidential TPU 8tx and memory-focused TPU 8i Max

3 sources · 3 articlesFirst reported Updated ()
What's new

November 10: TechCrunch adds an AMD design role, an April origin for the paired TPU 8i chips, a custom inferencing network, and Google’s claim of less reliance on outside cloud AI chips.

Merged summary · x-ai/grok-4.7 · rewritten as coverage arrivesSummaries by Vani AI

Google has unveiled eighth-generation TPU 8tx for confidential AI work and TPU 8i Max for higher memory and compute.

Google has added two eighth-generation Tensor Processing Unit variants, TPU 8tx and TPU 8i Max, to its cloud AI hardware lineup. The Decoder says this brings the family to four variants shipped in one year; TPU 8tx protects data during processing, while TPU 8i Max uses the same silicon in a configuration that doubles compute and memory in one rack. TechCrunch reports TPU 8tx was designed with AMD from TPU 8t for confidential training and inference, and that TPU 8i Max pairs two TPU 8i chips introduced in April, doubling memory, processing power and bandwidth. MarkTechPost’s headline calls TPU 8i Max a dual-chip TPU, but its retrieved page has no specifications. The Decoder also reports open-sourced TorchTPU, PyTorch, vLLM and SGLang support, and nearly 100 external TPU customers.

Key facts
  • The Decoder counts four eighth-generation TPU variants shipped in one year.
  • TechCrunch says TPU 8tx was designed with AMD and based on TPU 8t.
  • TechCrunch says TPU 8i Max pairs two April TPU 8i chips.
  • The Decoder reports nearly 100 external TPU customers.
  • The Decoder says TorchTPU was open sourced, with PyTorch, vLLM and SGLang support.
Explain this AI news with Vani AIWhat was announced, what is new and why it matters, in a new chat

What each outlet adds

beyond the shared facts
  • The DecoderFour variants, rack-level doubling, framework support and nearly 100 customers.
  • TechCrunchAMD partnership, paired April chips, bandwidth doubling and inferencing network.
  • MarkTechPostHeadline calls TPU 8i Max dual-chip, but page lacks specs.
Vendors
GoogleAMD
Products
TPU 8txTPU 8i MaxTPU 8tTPU 8iTPU
Organizations
MicrosoftAmazonOpenAIAnthropicNvidia

Coverage

oldest first · one row per article · UTC
  1. · 2d ago The DecoderFirst report

    Google Ships Four TPU Variants in One Year

    Google adds TPU 8tx and TPU 8i Max targeting confidential and memory-intensive AI workloads.

  2. · 2d ago MarkTechPost

    Google unveils dual-chip TPU 8i Max and confidential TPU 8tx

    The page body was only a cookie notice, so chip specs, availability and customers are not stated.

  3. · 1d ago TechCrunch · AI

    Google unveils TPU 8tx and TPU 8i Max for AI workloads

    TPU 8tx, designed with AMD, targets confidential training and inference; TPU 8i Max doubles memory in one package.