Google Expands Gen-8 Lineup With TPU 8tx, 8i Max: How This Targets AI Agent Workloads

Google has expanded its eighth-generation Tensor Processing Unit (TPU) portfolio with two new systems, TPU 8tx and TPU 8i Max, as it looks to capture a larger share of the growing enterprise artificial intelligence (AI) market.Reuters reported that the latest additions are designed to address two increasingly important requirements for businesses deploying AI agents: secure data processing and the ability to handle large volumes of information. The announcement also marks the first time Google has introduced four separate TPU systems in a single calendar year.The move comes as the Alphabet-owned company steps up its challenge to Nvidia&#039;s dominance in the AI chip market. Google&#039;s shares rose more than 3% in early trading following the announcement on Monday.ALSO READ:&amp;nbsp;Are Users Moving Away From Google? ChatGPT Search Volume Tells A StoryAccording to Google, the TPU 8tx is designed for government agencies and businesses operating in regulated sectors where protecting sensitive information is a priority. The system combines Google&#039;s TPU technology with processors from Advanced Micro Devices (AMD) and uses hardware-level encryption to protect data while it is being processed.Google said the 8tx can deliver up to twice the speed of its predecessor with virtually no performance trade-off, making it suitable for organisations that need secure AI inference without compromising computing efficiency.The TPU 8i Max, meanwhile, targets AI applications that require large context windows, greater memory capacity and higher computing power. By combining two chips into a single package, Google aims to support workloads such as coding assistants, mathematical reasoning and drug discovery.These capabilities are becoming increasingly relevant as businesses move beyond conventional AI chatbots towards autonomous AI agents that can process information, reason through complex tasks and execute multiple actions.Google said its TPU business now serves nearly 100 external customers, including financial institutions, research organisations, universities and retailers. These customers use the systems for AI model training, inference, agentic workflows and deployments within their own data centres.The latest launch follows the introduction of the TPU 8t and TPU 8i earlier this year. While the 8t was developed primarily for AI training, offering up to 2.7 times the performance of previous generations, the 8i focused on real-time inference, delivering an 80% improvement in price-performance, according to Google.ALSO READ:&amp;nbsp;Alphabet Shares Extend Decline To 13.7%, Biggest Drop On RecordThe company said its approach centres on designing complete computing systems rather than focusing solely on chip performance. Its AI Hypercomputer architecture combines processing power with memory, optical networking and security features to address specific customer requirements.Google also highlighted the role of its internal AI tools, including AlphaChip and Gemini models, in accelerating chip design. These systems help engineers automate physical chip layouts, reducing processes that previously took weeks to just hours.The expansion comes as demand for AI infrastructure drives growth at Google Cloud. Alphabet reported that Google Cloud revenue more than doubled to $30.71 billion in the September quarter, with TPU sales making a meaningful contribution to the results.

Google Expands Gen-8 Lineup With TPU 8tx, 8i Max: How This Targets AI Agent Workloads

Google's latest AI chips target secure computing and memory-intensive workloads as businesses increasingly deploy autonomous AI agents, strengthening its push against Nvidia.

News for You

India Vs New Zealand Live Streaming: When And Where To Watch IND Vs NZ 3rd ODI

'Maharshi Vasooli': THIS Ramayana Meme Has Golmaal Connection And Fans Can't Keep Calm

FDA Adds Warning Label To Obesity Drugs For Eye Condition

Japan Masters 2026: Schedule, Live Streaming, Indian Shuttlers In Action And Prize Money