TPU 8tx and TPU 8i Max

Google has expanded its eighth-generation TPU portfolio with two new systems: TPU 8tx for secure and confidential inference, and TPU 8i Max for massive context windows and high compute density per rack. The TPU 8tx provides specialized hardware for security and density, while the TPU 8i Max provides high compute density per rack. This expansion comes after introducing two new specialized TPUs in the same generation at Google Cloud Next in April, which offered up to 2.7 times better training performance and 80% better inference price performance. The eighth-generation TPU portfolio now includes four systems: TPU 8t, TPU 8i, TPU 8tx, and TPU 8i Max. Google designed TPUs over a decade ago to meet its internal demands, including voice recognition and language translation, and has since expanded its use cases to training and serving its own Gemini models, agentic workloads, and delivering services to Google Cloud customers. TPUs are now a rapidly growing business with nearly 100 customers. The new systems are designed to help manage complex workloads and provide specialized hardware for security, density, and reasoning. This expansion is a continuation of Google's efforts to improve its cloud infrastructure and provide more options for its customers.