AI CURATED HOT STORY · 11/9/2026
Google has expanded its custom TPU lineup with two new variants: TPU 8tx for confidential computing and TPU 8i Max for ultra-long context scenarios, bringing the total to four products. Both are reconfigured from existing designs, with 8tx working alongside AMD chips. Google's chip chief said the iteration pace has accelerated from one chip every two years to multiple per year, aiming to grab AI chip market share amid Nvidia supply constraints.
TPU 8tx keeps data encrypted during processing for government and regulated industries, while TPU 8i Max doubles memory and compute per rack via dual-chip packaging for drug discovery, math, and coding applications. With software tools like TorchTPU, Google is lowering barriers for developers and pushing its chips to external customers for inference workloads.
The expansion not only broadens TPU use cases but also intensifies competition with Nvidia through faster iteration and supply chain partnerships like AMD. As inference demand grows, Google could attract more cloud customers with cost-performance and customization advantages, potentially reshaping the AI hardware landscape.
The line compares only sources observed throughout; gap hours are interpolated to keep the trend continuous.