Google releases two TPU variants, the 8tx and 8i Max, targeting confidential computing and long-context applications respectively

Cailianshe Financial News Agency

Open the APP

Google releases TPU 8tx and 8i Max to accelerate the challenge to Nvidia's dominance in AI chips

Cailianshe Li Ying, Editor Zhang Xiaolong 2026-11-09 22:14

① Google releases two variants, TPU 8tx and 8i Max, targeting confidential computing and long-context AI applications respectively; ② Research institutions expect TPU sales to grow from about $12 billion this year to more than $135 billion by 2028; ③ Anthropic has the right to purchase up to $150 billion of TPUs between 2028 and 2031.

Cailianshe, November 9 (Editor Li Ying) Google on Monday local time released two new variants of its eighth-generation TPU, targeting confidential computing and long-context AI applications respectively, further challenging Nvidia by segmenting its product line. Demand for TPUs is strong and supply has already been fully booked, but component shortages are constraining supply, and the risk of a slowdown in AI infrastructure spending has not dissipated.

According to media reports on November 9 local time, Google that day released two new variants of its Tensor Processing Unit (TPU): TPU 8tx and 8i Max. TPU is Google's self-developed series of AI-specific chips, used for training and inference of AI models, competing with Nvidia's GPUs.

Google's chip division has historically launched a new generation of TPU roughly every two years. According to information on Google's official website, on April 22 this year the company simultaneously released two chips of its eighth-generation TPU at the Google Cloud Next conference, namely the TPU 8t focused on model training and the TPU 8i focused on inference.

The 8tx and 8i Max are derivatives of these two chips, expanding the eighth-generation TPU to four models and markedly accelerating the development pace.

Amin Vahdat, head of Google's chip business, said in an interview that these four products will each meet different needs. He noted:

"This is consistent with Google's long-standing judgment that the future belongs to specialization, and the benefits of specialization remain very significant."

"We could build one chip that tries to cover the entire market, but launching specialized versions can create much greater value."

Vahdat said the new products are not entirely new chip designs but products reconfigured according to customer needs. Among them, the 8tx targets the confidential computing needs of government agencies and enterprises in regulated industries, working in concert with AMD's central processing units (CPUs), and can encrypt data at the chip level, avoiding data being stored in memory in decrypted form during processing.

The 8i Max integrates two TPU chips in the same package, providing stronger computing power and larger memory, mainly serving AI applications that require long context windows, such as drug discovery, mathematics and programming scenarios.

Vahdat disclosed that the two new products will go on sale next year.

Segmenting chip products by task is gradually becoming an industry trend. Even Nvidia, which has long positioned GPUs as general-purpose AI products, has begun offering chip systems for specific tasks, including launching its own CPUs and, using technology obtained from AI chip startup Groq, launching chips focused on the inference stage.

Commercialization accelerates

Google began developing TPUs more than a decade ago in cooperation with chip design company Broadcom, initially to meet its own growing computing demand more efficiently, and has now begun selling them externally beyond its cloud business.

According to industry research institutions, by 2028 TPU sales, including related software and hardware, will exceed $135 billion, with this year's estimate at about $12 billion. This figure does not yet include revenue generated by customers using TPUs through Google Cloud.

Institutional analyst Mandeep Singh noted that the appeal of TPUs lies in cost-effectiveness: prices are about 30% lower than Nvidia GPUs, while throughput is quite competitive.

Currently, tech giants such as Apple, Meta and Oracle are all TPU customers.

Oracle announced last month that it has signed an agreement with Google allowing its cloud customers to use TPUs for the first time starting next year. Before that, AI chip options for Oracle cloud customers were limited to products from Nvidia and AMD.

AI startup Anthropic is also a TPU customer. According to a recent regulatory filing, the company has the right to purchase up to $150 billion of TPUs between 2028 and 2031. According to media reports, ChatGPT developer OpenAI also plans to procure TPUs.

Google disclosed in its earnings report last month that quantitative trading firm Jump Trading is also a customer. The company's chief technology officer, Alex Davies, said the company has deployed thousands of TPUs to train AI models for trading research and other purposes, but training work still mainly relies on Nvidia chips; in areas where TPUs perform no worse than Nvidia chips, their higher energy efficiency also helps reduce costs.

Challenges remain

Amid surging demand, Google also faces competition from new entrants. According to reports, chip startup Tenspresso, co-founded by two former TPU engineers, is also targeting long-context inference with its first chip, but it will take at least another year to complete the design. OpenAI and Anthropic are also advancing their own chips, with teams that include many veteran TPU engineers.

Institutional analyst Singh expects that opposition to data centers could cut tech companies' originally projected $1.4 trillion in spending for 2027 by about 5% to 7%. He also noted that Google's TPU profit margins are lower than Nvidia's.

Like the entire industry, Google also faces shortages of memory chips and manufacturing capacity, making it difficult to fully meet surging demand. Google said in its earnings report that all of its available TPUs have been booked by customers. Vahdat's team is diversifying supply sources as much as possible.

Media reports noted that in recent months, Google has sought partners beyond Broadcom, its main TPU partner, signing an agreement with Broadcom competitor and chipmaker Marvell Technology covering part of its TPU business, though disclosed details are limited; other reports say chip design company MediaTek is also involved in Google's TPU business.

Research analyst William Tong believes Google is likely finalizing longer-term agreements to lock in supply and avoid further bottlenecks.

Global Market Intelligence

Artificial Intelligence

Cailian Press statement: The content of this article is for reference only and does not constitute investment advice. Investors who act on it do so at their own risk.

Business Cooperation

Popular Unlocks

[Telegram Interpretation] Doubling down on competition with Nvidia, Google launches two new TPU chips; East China's first domestic TPU thousand-card intelligent computing cluster has been completed, and a company in which this company holds a stake provides TPU products for the cluster

Doubling down on competition with Nvidia, Google launches two new TPU chips; East China's first domestic TPU thousand-card intelligent computing cluster has been completed, and a company in which this company holds a stake provides TPU products for the cluster, while another invested company's products can be used for packaging high-computing-power chips such as TPU/CPU/GPU chips.

[All About Announcements] Humanoid robots + embodied intelligence + machine vision + high-end equipment! This company's industrial robot shipments have ranked first for seven consecutive quarters

① Humanoid robots + embodied intelligence + machine vision + high-end equipment! This company's industrial robot shipments have ranked first for seven consecutive quarters; ② Advanced packaging + photomasks + third-generation semiconductors! This company plans to invest 1.5 billion yuan to step up high-end semiconductor photomasks; ③ MLCC + optical modules + commercial aerospace + Huawei! The company plans to raise no more than 1 billion yuan through a private placement to build ceramic thin-film circuits and other projects.

[Trend Research Report · Insight] Shingled technology becomes a key link in "space photovoltaics," significantly increasing the effective power-generating area for the same solar wing area; the competitive landscape in equipment and adhesive links is now preliminarily clear, with leading positioned players enjoying obvious first-mover advantages; AI may usher in a narrative window of "rise in the East and decline in the West"

① AI may usher in a narrative window of "rise in the East and decline in the West"; ② Shingled technology becomes a key link in "space photovoltaics," significantly increasing the effective power-generating area for the same solar wing area; the competitive landscape in equipment and adhesive links is now preliminarily clear, with leading positioned players enjoying obvious first-mover advantages; ③ Today, institutions across the market published a total of 328 research reports, Yealink Network's rating was upgraded, and 20 companies received first-time coverage, among which Hangcha Group received in-depth coverage from a New Fortune analyst; ④ In the ranking of institutional attention to individual stocks, Ecovacs made the list for the first time, with the top five in order being Roborock > Haitian Flavouring > Anjoy Foods > Dongpeng Beverage > Ecovacs.

U.S. stock index futures all rise premarket; Nvidia plunges 6% against the market | Tonight's highlights

Cailian Press Zhao Hao 2026-11-09 21:04

① Nvidia's main competitor AMD rises more than 4%; ② U.S. Democrats plan to launch investigations into Musk and others; ③ onsemi terminates its $2 billion SiC expansion plan in the Czech Republic; ④ Terafab target exposed: aiming for monthly production of 700,000 memory chip wafers.

Tesla Cybertruck to be discontinued next year; Musk says deposits will be refunded to a large number of customers

Cailian Press Niu Zhanlin 2026-11-09 20:47

① Tesla will shut down the Cybertruck production line in about eight months, at which point it will have to refund deposits to a large number of customers; ② Musk said the complexity of maintaining Cybertruck production carries an opportunity cost far exceeding the value that could be created by increasing Cybercab output.