{"content_id":"5rcsar9m5d","slug":"google-tpu-8tx-8i-max-announcement","locale":"en","schema_type":"NewsArticle","category":"news","category_name":"News","title":"Google TPU 8tx and 8i Max announced, launch date undisclosed","summary":"Google announced that it has added TPU 8tx and TPU 8i Max to its 8th-generation TPU lineup. The official information does not confirm the actual availability date, availability in Korea, or pricing and quota terms.","sponsorship_disclosure":null,"affiliate_disclosure":null,"commerce_disclosure":null,"author":{"name":"Injoys Editorial Team","url":"https://injoys.com/ko/about"},"key_points":["Google announced TPU 8tx and TPU 8i Max on November 9, 2026.","TPU 8i Max is a system that combines two low-latency TPU 8i chips in a single package.","Google Cloud said TPU 8tx is optimized for secure and confidential inference.","Google Cloud said TPU 8i Max has twice the compute performance, SRAM and HBM bandwidth of TPU 8i.","The actual availability date, availability in Korea and the Seoul region, and pricing and contract terms remain unconfirmed."],"content_markdown":"Google announced TPU 8tx and 8i Max on November 9, 2026. The announcement was made at 13:00 UTC, or 22:00 in South Korea on the same day. The official description does not confirm who can currently use them or when access actually begins.\n\nThis article follows Google's official description and is current as of November 10, 2026.\n\n## 8tx and 8i Max added to the eighth-generation TPU lineup\n\nGoogle said it added two systems to its eighth-generation TPU lineup. TPU 8i Max combines two low-latency TPU 8i chips in a single package. Google Cloud described TPU 8tx as optimized for secure and confidential inference.\n\n| Product | Features stated by the company |\n| --- | --- |\n| TPU 8tx | Optimized for secure and confidential inference |\n| TPU 8i Max | Combines two low-latency TPU 8i chips in a single package |\n\n## Availability in South Korea and access start date not disclosed\n\nThe actual access start date and availability in South Korea or the Seoul region are unconfirmed. The official description also does not specify which customers can currently use the systems. The product announcement alone is not enough to conclude that they are immediately available in South Korea.\n\nAPI identifiers for the new systems are unconfirmed. The published description also does not explain how to provision them by creating resources and having them allocated. This additional information is needed to connect them to development environments in South Korea.\n\n## Pricing and minimum contract and quota requirements not disclosed\n\nPricing and minimum contract and quota requirements for the two systems are unconfirmed. As a result, the published specifications alone are not enough to calculate actual adoption costs. Contract or cost reviews require official confirmation of the following conditions.\n\n- Pricing and billing criteria for each system\n- Minimum contract requirements\n- Resource quotas and allocation conditions\n\n## 8i Max's twofold figures come from the company\n\nGoogle Cloud compared TPU 8i Max with TPU 8i in three areas. It stated that compute performance, SRAM, and HBM bandwidth are each **twice** those of TPU 8i. The figures below come from the company's description and were not presented as independent measurements.\n\n| Comparison item | TPU 8i Max relative to TPU 8i | Source of figures |\n| --- | --- | --- |\n| Compute performance | 2x | Google Cloud description |\n| SRAM | 2x | Google Cloud description |\n| HBM bandwidth | 2x | Google Cloud description |\n\nThese multiples are comparisons of individual specifications. They cannot be taken to mean that overall workloads run twice as fast. Independent test results from Artificial Analysis for the two systems could not be confirmed.\n\n## Memory capacity and supported context length also unconfirmed\n\nThe exact memory capacity and supported context length are unconfirmed. The relative SRAM multiple alone does not reveal its absolute capacity. The HBM bandwidth figure does not indicate memory capacity either.\n\n- Memory capacity: Exact capacity figures unconfirmed\n- Supported context length: Input length that can be processed unconfirmed\n- Independent performance evaluation: Independent test results for the two new systems unconfirmed\n\n## Conditions to check before adoption in South Korea\n\nWhen considering adoption in South Korea, first check the available regions and actual terms of access. Memory capacity and input length needed for the work must also be checked. The specification multiples provided by the company should be assessed separately from actual workload performance.\n\n| Review item | Details to check in official information |\n| --- | --- |\n| Availability | Access start date, eligible users, availability in South Korea or the Seoul region |\n| Costs and contracts | Pricing, billing criteria, minimum contract and quota requirements |\n| Development environment integration | API identifiers and provisioning methods |\n| Workload suitability | Exact memory capacity and supported context length |\n| Performance evidence | Measurement conditions for the company's figures and separate independent test results |","content_html":"\u003cp\u003eGoogle announced TPU 8tx and 8i Max on November 9, 2026. The announcement was made at 13:00 UTC, or 22:00 in South Korea on the same day. The official description does not confirm who can currently use them or when access actually begins.\u003c/p\u003e\n\u003cp\u003eThis article follows Google's official description and is current as of November 10, 2026.\u003c/p\u003e\n\u003ch2\u003e\n\u003ca href=\"#8tx-and-8i-max-added-to-the-eighth-generation-tpu-lineup\" class=\"anchor\" id=\"8tx-and-8i-max-added-to-the-eighth-generation-tpu-lineup\"\u003e\u003c/a\u003e8tx and 8i Max added to the eighth-generation TPU lineup\u003c/h2\u003e\n\u003cp\u003eGoogle said it added two systems to its eighth-generation TPU lineup. TPU 8i Max combines two low-latency TPU 8i chips in a single package. Google Cloud described TPU 8tx as optimized for secure and confidential inference.\u003c/p\u003e\n\u003cdiv class=\"overflow-x-auto\"\u003e\u003ctable\u003e\n\u003cthead\u003e\n\u003ctr\u003e\n\u003cth\u003eProduct\u003c/th\u003e\n\u003cth\u003eFeatures stated by the company\u003c/th\u003e\n\u003c/tr\u003e\n\u003c/thead\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Product\"\u003eTPU 8tx\u003c/td\u003e\n\u003ctd data-label=\"Features stated by the company\"\u003eOptimized for secure and confidential inference\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Product\"\u003eTPU 8i Max\u003c/td\u003e\n\u003ctd data-label=\"Features stated by the company\"\u003eCombines two low-latency TPU 8i chips in a single package\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003c/table\u003e\u003c/div\u003e\n\u003ch2\u003e\n\u003ca href=\"#availability-in-south-korea-and-access-start-date-not-disclosed\" class=\"anchor\" id=\"availability-in-south-korea-and-access-start-date-not-disclosed\"\u003e\u003c/a\u003eAvailability in South Korea and access start date not disclosed\u003c/h2\u003e\n\u003cp\u003eThe actual access start date and availability in South Korea or the Seoul region are unconfirmed. The official description also does not specify which customers can currently use the systems. The product announcement alone is not enough to conclude that they are immediately available in South Korea.\u003c/p\u003e\n\u003cp\u003eAPI identifiers for the new systems are unconfirmed. The published description also does not explain how to provision them by creating resources and having them allocated. This additional information is needed to connect them to development environments in South Korea.\u003c/p\u003e\n\u003ch2\u003e\n\u003ca href=\"#pricing-and-minimum-contract-and-quota-requirements-not-disclosed\" class=\"anchor\" id=\"pricing-and-minimum-contract-and-quota-requirements-not-disclosed\"\u003e\u003c/a\u003ePricing and minimum contract and quota requirements not disclosed\u003c/h2\u003e\n\u003cp\u003ePricing and minimum contract and quota requirements for the two systems are unconfirmed. As a result, the published specifications alone are not enough to calculate actual adoption costs. Contract or cost reviews require official confirmation of the following conditions.\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003ePricing and billing criteria for each system\u003c/li\u003e\n\u003cli\u003eMinimum contract requirements\u003c/li\u003e\n\u003cli\u003eResource quotas and allocation conditions\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch2\u003e\n\u003ca href=\"#8i-maxs-twofold-figures-come-from-the-company\" class=\"anchor\" id=\"8i-maxs-twofold-figures-come-from-the-company\"\u003e\u003c/a\u003e8i Max's twofold figures come from the company\u003c/h2\u003e\n\u003cp\u003eGoogle Cloud compared TPU 8i Max with TPU 8i in three areas. It stated that compute performance, SRAM, and HBM bandwidth are each \u003cstrong\u003etwice\u003c/strong\u003e those of TPU 8i. The figures below come from the company's description and were not presented as independent measurements.\u003c/p\u003e\n\u003cdiv class=\"overflow-x-auto\"\u003e\u003ctable\u003e\n\u003cthead\u003e\n\u003ctr\u003e\n\u003cth\u003eComparison item\u003c/th\u003e\n\u003cth\u003eTPU 8i Max relative to TPU 8i\u003c/th\u003e\n\u003cth\u003eSource of figures\u003c/th\u003e\n\u003c/tr\u003e\n\u003c/thead\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Comparison item\"\u003eCompute performance\u003c/td\u003e\n\u003ctd data-label=\"TPU 8i Max relative to TPU 8i\"\u003e2x\u003c/td\u003e\n\u003ctd data-label=\"Source of figures\"\u003eGoogle Cloud description\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Comparison item\"\u003eSRAM\u003c/td\u003e\n\u003ctd data-label=\"TPU 8i Max relative to TPU 8i\"\u003e2x\u003c/td\u003e\n\u003ctd data-label=\"Source of figures\"\u003eGoogle Cloud description\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Comparison item\"\u003eHBM bandwidth\u003c/td\u003e\n\u003ctd data-label=\"TPU 8i Max relative to TPU 8i\"\u003e2x\u003c/td\u003e\n\u003ctd data-label=\"Source of figures\"\u003eGoogle Cloud description\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003c/table\u003e\u003c/div\u003e\n\u003cp\u003eThese multiples are comparisons of individual specifications. They cannot be taken to mean that overall workloads run twice as fast. Independent test results from Artificial Analysis for the two systems could not be confirmed.\u003c/p\u003e\n\u003ch2\u003e\n\u003ca href=\"#memory-capacity-and-supported-context-length-also-unconfirmed\" class=\"anchor\" id=\"memory-capacity-and-supported-context-length-also-unconfirmed\"\u003e\u003c/a\u003eMemory capacity and supported context length also unconfirmed\u003c/h2\u003e\n\u003cp\u003eThe exact memory capacity and supported context length are unconfirmed. The relative SRAM multiple alone does not reveal its absolute capacity. The HBM bandwidth figure does not indicate memory capacity either.\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eMemory capacity: Exact capacity figures unconfirmed\u003c/li\u003e\n\u003cli\u003eSupported context length: Input length that can be processed unconfirmed\u003c/li\u003e\n\u003cli\u003eIndependent performance evaluation: Independent test results for the two new systems unconfirmed\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch2\u003e\n\u003ca href=\"#conditions-to-check-before-adoption-in-south-korea\" class=\"anchor\" id=\"conditions-to-check-before-adoption-in-south-korea\"\u003e\u003c/a\u003eConditions to check before adoption in South Korea\u003c/h2\u003e\n\u003cp\u003eWhen considering adoption in South Korea, first check the available regions and actual terms of access. Memory capacity and input length needed for the work must also be checked. The specification multiples provided by the company should be assessed separately from actual workload performance.\u003c/p\u003e\n\u003cdiv class=\"overflow-x-auto\"\u003e\u003ctable\u003e\n\u003cthead\u003e\n\u003ctr\u003e\n\u003cth\u003eReview item\u003c/th\u003e\n\u003cth\u003eDetails to check in official information\u003c/th\u003e\n\u003c/tr\u003e\n\u003c/thead\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Review item\"\u003eAvailability\u003c/td\u003e\n\u003ctd data-label=\"Details to check in official information\"\u003eAccess start date, eligible users, availability in South Korea or the Seoul region\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Review item\"\u003eCosts and contracts\u003c/td\u003e\n\u003ctd data-label=\"Details to check in official information\"\u003ePricing, billing criteria, minimum contract and quota requirements\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Review item\"\u003eDevelopment environment integration\u003c/td\u003e\n\u003ctd data-label=\"Details to check in official information\"\u003eAPI identifiers and provisioning methods\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Review item\"\u003eWorkload suitability\u003c/td\u003e\n\u003ctd data-label=\"Details to check in official information\"\u003eExact memory capacity and supported context length\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Review item\"\u003ePerformance evidence\u003c/td\u003e\n\u003ctd data-label=\"Details to check in official information\"\u003eMeasurement conditions for the company's figures and separate independent test results\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003c/table\u003e\u003c/div\u003e\n","tags":["AI Data Center","AI Development","AI chips","HBM"],"faqs":[{"question":"Can TPU 8tx and TPU 8i Max be used right away in Korea?","answer":"The actual availability date and who can currently use them have not been confirmed. Availability in Korea and the Seoul region is also not specified in the official description."},{"question":"Have pricing and minimum contract requirements for the two systems been disclosed?","answer":"Pricing and minimum contract and quota requirements for the two systems have not been confirmed. The specifications currently available are not enough to calculate the cost of adoption."},{"question":"Can developers in Korea find information on which API to connect to and how to configure it?","answer":"The API identifiers and provisioning methods for the new systems have not been confirmed. To connect them to a development environment, you need to check whether that information has been officially released."}],"sources":[{"url":"https://blog.google/innovation-and-ai/infrastructure-and-cloud/global-network/tpu-8i-max-and-8tx/","title":"Our newest TPUs: 8tx and 8i Max, built for the agentic era - Google","type":"source"},{"url":"https://cloud.google.com/tpu","title":"Tensor Processing Units (TPUs) | Google Cloud - Google Cloud","type":"source"}],"images":[{"id":2647,"url":"https://injoys.com/rails/active_storage/blobs/proxy/eyJfcmFpbHMiOnsiZGF0YSI6NDA4MDcsInB1ciI6ImJsb2JfaWQifX0=--5e8809affc745dfc3329226e3e8f5761be96ab80/ai-5afc843a.webp","is_representative":true,"generation_method":"ai_photo","license":"ai_generated","mime_type":"image/webp","width":1536,"height":1024,"translations":{"ko":{"alt":"서버 랙 앞에서 여성이 열린 서버 보드 위 두 개의 반도체 다이를 살펴보고 있다.","caption":"Google은 TPU 8i Max의 연산 성능과 HBM 대역폭이 TPU 8i의 두 배라고 밝혔다.","description":null},"en":{"alt":"A woman examines two silicon dies on an exposed server board in front of server racks.","caption":"Google says TPU 8i Max doubles TPU 8i’s compute performance and HBM bandwidth.","description":null},"ja":{"alt":"サーバーラックの前で、女性が露出したサーバー基板上の2つの半導体ダイを見つめている。","caption":"Googleによると、TPU 8i Maxの演算性能とHBM帯域幅はTPU 8iの2倍となる。","description":null},"es":{"alt":"Una mujer examina dos dados de silicio en una placa de servidor expuesta, frente a varios racks.","caption":"Según Google, TPU 8i Max duplica la capacidad de cómputo y el ancho de banda HBM de TPU 8i.","description":null},"id":{"alt":"Seorang perempuan mengamati dua die silikon pada papan server terbuka di depan rak server.","caption":"Menurut Google, performa komputasi dan bandwidth HBM TPU 8i Max dua kali lipat dari TPU 8i.","description":null},"pt":{"alt":"Uma mulher examina dois dies de silício em uma placa de servidor exposta, diante de racks.","caption":"Segundo o Google, o TPU 8i Max oferece o dobro do desempenho computacional e da largura de banda HBM do TPU 8i.","description":null},"zh-hant":{"alt":"一名女子在伺服器機櫃前，仔細查看開放式伺服器主板上的兩顆晶粒。","caption":"Google表示，TPU 8i Max的運算效能與HBM頻寬皆為TPU 8i的兩倍。","description":null},"de":{"alt":"Eine Frau betrachtet zwei Silizium-Dies auf einer freigelegten Serverplatine vor Serverracks.","caption":"Laut Google verdoppelt TPU 8i Max die Rechenleistung und HBM-Bandbreite gegenüber TPU 8i.","description":null}}},{"id":2648,"url":"https://injoys.com/rails/active_storage/blobs/proxy/eyJfcmFpbHMiOnsiZGF0YSI6NDA4MTMsInB1ciI6ImJsb2JfaWQifX0=--e5121dad8573e283e6590a2aa1c00cf1df776362/ai-5ae217fd.webp","is_representative":false,"generation_method":"ai_semi","license":"ai_generated","mime_type":"image/webp","width":1536,"height":1024,"translations":{"ko":{"alt":"여성 엔지니어가 잠금장치가 달린 서버 캐비닛을 살펴보고 있으며, 옆에 방패와 자물쇠 기호가 떠 있다.","caption":"Google Cloud는 TPU 8tx가 보안·기밀 추론에 최적화됐다고 설명했다.","description":null},"en":{"alt":"A female engineer inspects a locked server cabinet beside an overlaid shield and padlock symbol.","caption":"Google Cloud says TPU 8tx is optimized for secure and confidential inference.","description":null},"ja":{"alt":"女性エンジニアが施錠されたサーバーキャビネットを確認し、その横に盾と南京錠のマークが重なっている。","caption":"Google Cloudは、TPU 8txがセキュアかつ機密性の高い推論に最適化されていると説明した。","description":null},"es":{"alt":"Una ingeniera revisa un armario de servidores cerrado, junto a un símbolo superpuesto de escudo y candado.","caption":"Google Cloud afirma que TPU 8tx está optimizado para la inferencia segura y confidencial.","description":null},"id":{"alt":"Seorang insinyur perempuan memeriksa lemari server terkunci, dengan simbol perisai dan gembok di sampingnya.","caption":"Google Cloud menyatakan TPU 8tx dioptimalkan untuk inferensi yang aman dan rahasia.","description":null},"pt":{"alt":"Uma engenheira inspeciona um gabinete de servidores trancado, com um símbolo sobreposto de escudo e cadeado.","caption":"O Google Cloud afirma que o TPU 8tx é otimizado para inferência segura e confidencial.","description":null},"zh-hant":{"alt":"一名女性工程師查看上鎖的伺服器機櫃，旁邊疊有盾牌與掛鎖符號。","caption":"Google Cloud表示，TPU 8tx針對安全與機密推論進行了最佳化。","description":null},"de":{"alt":"Eine Ingenieurin prüft einen verriegelten Serverschrank neben einem eingeblendeten Schild- und Schlosssymbol.","caption":"Laut Google Cloud ist TPU 8tx für sichere und vertrauliche Inferenz optimiert.","description":null}}}],"published_at":"2026-11-10T07:43:40+09:00","updated_at":"2026-11-10T07:43:40+09:00","license":"cc_by","translation_status":"reviewed","available_locales":["ko","en","ja","es"],"data_locales":["ko","en","ja","es","id","pt","zh-hant","de"],"url":"https://injoys.com/en/articles/google-tpu-8tx-8i-max-announcement"}