🔍 Read the full analysis: What Happens When Google Builds More AI Chips In One Year? on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Google announced two new versions of its eighth-generation AI chips on Nov. 9, bringing the number of TPU systems it introduced this year to four. The TPU 8tx is designed for confidential computing and the TPU 8i Max for long-context workloads, but both are still listed as coming soon and are expected in 2027. Their performance figures are Google’s claims; prices, cloud regions and independent benchmarks have not been published.
Google announced two new AI chip systems on Nov. 9, bringing its total number of TPU systems introduced in 2026 to four and broadening its alternatives to Nvidia GPUs. The TPU 8tx is aimed at confidential computing, while the TPU 8i Max is designed for workloads with very long inputs; neither is available yet, and Google has said the new variants are due in 2027.
The additions extend the eighth-generation TPU lineup Google first introduced at its Cloud Next conference on April 22. That announcement included the TPU 8t, designed for training AI models, and the TPU 8i, designed for inference, the stage when trained models generate responses. The November systems adapt that platform for more specific needs rather than introducing entirely new silicon designs.
Google says the TPU 8tx uses the same silicon as the TPU 8t and is paired with fifth-generation AMD EPYC processors. The company says it is built to keep sensitive data and model weights protected while in use, and claims up to two times lower latency than the TPU 8t with a near-zero performance penalty. That figure has not been independently tested.
The TPU 8i Max combines two TPU 8i chips on one package. Google says the configuration doubles the base TPU 8i’s compute, on-chip SRAM and high-bandwidth memory bandwidth, while fitting within the existing rack footprint. The company is targeting tasks such as processing large codebases or lengthy video. Google has not published absolute performance figures for either new system.
Google expands its AI chip lineup Four TPU systems in one year
Google introduced two specialized TPU systems in November. They broaden its cloud hardware options, but availability and real-world performance remain unproven.
Two systems, distinct workloads
The additions tailor the eighth-generation platform to specific computing needs.
TPU 8tx
Designed to protect sensitive data and model weights while in use. Google claims up to two times lower latency than TPU 8t.
TPU 8i Max
Combines two TPU 8i chips on one package for workloads such as large codebases and lengthy video.
A broader cloud option
More specialized hardware could give businesses another supplier choice for developing and running AI models.
- Options for confidential workloads and long inputs
- Oracle plans to offer Google TPUs in 2027
- Moving from GPUs may require software and engineering work
- Cost and workload performance are not yet established
What remains unknown
Google has announced plans and vendor claims; customers still need details to assess fit.
The new systems are listed as coming soon, with availability expected in 2027.
Prices, exact launch dates and cloud regions have not been announced.
No independent benchmarks for either new system are provided.
A Broader Alternative to Nvidia GPUs
The announcement matters because access to AI computing capacity can shape how quickly companies develop and run models. Google is offering customers more specialized options within its own cloud infrastructure, including systems tailored to confidential workloads and to models that need to handle long inputs. A second supplier may give buyers more choice, though it does not by itself establish that Google’s systems are cheaper or faster for a particular workload.
Google says nearly 100 customers use its TPUs, and Alphabet has described a supply-constrained environment for AI computing. Oracle has also said it will offer Google TPUs in its cloud starting in calendar year 2027. Those developments could broaden access beyond Google Cloud, but customers considering a move from GPUs still face engineering work to adapt software and operating practices. The new systems’ real-world performance, pricing and availability will determine how useful that choice becomes.
Popular right now
Breville Barista Touch Impress Espresso Machine, Noir (BES881NRE)
De'Longhi Rivelia Automatic Espresso Machine with Grinder and Milk Frother, Beige
Vitamix VX1 Professional-Grade Blender, 64 oz, Black
As an affiliate, we earn on qualifying purchases.
From Two TPU Systems to Four
Google introduced two distinct eighth-generation systems in April: the TPU 8t for model training and the TPU 8i for inference. The company reported performance-per-dollar gains over the previous generation, Ironwood, of up to 2.7 times for TPU 8t in large-scale training and up to 80% for TPU 8i in inference. These are vendor comparisons, not independent test results.
Google says it has now introduced four TPU systems in a single year for the first time. Bloomberg reported that the company previously released a new AI processor about every two years. Google’s AI and infrastructure chief, Amin Vahdat, told Bloomberg the latest versions are not completely new silicon designs; they reconfigure existing silicon for tasks customers requested. The approach lets Google differentiate systems through packaging, memory, networking and security as well as chip design.
The company is also working on software support. Google points to TorchTPU, which it says lets PyTorch code and Hugging Face models run natively on TPUs, and TPU backends for the open-source inference tools vLLM and SGLang. Software compatibility is relevant to adoption, but the announcement does not establish how much effort customers will need to move existing workloads.
“This is the first time we have introduced four TPU systems in a single year.”
— Google, in its Nov. 9 announcement
Pricing and Independent Tests Are Missing
The TPU product page lists all four eighth-generation systems as coming soon. Vahdat told Bloomberg the two latest variants would be available in 2027, but Google has not announced exact launch dates, prices or cloud regions. The announcement also provides no independent benchmarks for the TPU 8tx or TPU 8i Max, and Google has published only relative performance claims for those systems.
It is also unclear how widely customers will adopt the chips or how they will perform across different models and workloads. Google’s figures for the base TPU 8t and 8i compare them with the previous generation under specified uses; they do not settle how the systems compare with Nvidia GPUs or other hardware in buyers’ own environments. The cost and engineering effort of adapting software to TPUs remain relevant unknowns.
Availability Is Expected in 2027
The next concrete milestone is Google’s planned 2027 availability for the TPU 8tx and TPU 8i Max. Buyers will need further details on launch timing, regions, pricing and performance to judge whether the systems fit their needs. Oracle’s planned offering of Google TPUs in its cloud, also starting in calendar year 2027, may give customers another route to access the hardware.
Until products are available and tested outside Google, the announcement establishes the company’s intended chip lineup and its own performance claims—not demonstrated results in customer workloads. Future specifications, independent testing and customer deployments should clarify whether the specialized systems deliver practical benefits over existing options.
Key Questions
What did Google announce?
Google announced the TPU 8tx for confidential computing and the TPU 8i Max for workloads involving very long inputs. They bring the number of TPU systems Google introduced in 2026 to four.
When will the new TPUs be available?
Google’s Amin Vahdat told Bloomberg that the two new variants are expected in 2027. Exact launch dates and cloud regions have not been announced.
Are the performance claims independently verified?
No independent benchmarks for the TPU 8tx or TPU 8i Max were provided in the source material. Figures such as the claimed twofold latency improvement for the 8tx are Google’s claims.
How do the new systems differ?
The TPU 8tx is intended to protect data and model weights while they are being processed. The TPU 8i Max packages two TPU 8i chips together; Google says it doubles the base system’s compute, SRAM and memory bandwidth.
Does the announcement mean Google is replacing Nvidia?
No. It gives customers another potential AI-computing option, but the systems are not yet available and their comparative performance, pricing and adoption are unknown. Moving workloads to TPUs may also require software and engineering changes.
Source: ThorstenMeyerAI.com
