Google Cloud released two new versions of its TPU AI processors Monday, rounding out the eighth generation of its homegrown challenger to Nvidia's control of the AI chip market.
The new TPUs complement the 8t and 8i chips launched earlier this year; the "t" stands for model training, and the "i" targets inference workloads. The new 8tx includes one of AMD's Epyc CPUs to create a secure hardware processing enclave, and the 8iMax was designed to expand the amount of memory that the chip can handle to reduce a key AI bottleneck.
"This is the first time we have introduced four TPU systems in a single year," Google Cloud's Amin Vahdat said in a blog post. "With computing moving from human to agent scale, businesses are asking for accelerators that can reason across AI models’ massive context windows. In addition, some highly regulated customers are asking for specialized encryption capabilities for unique workloads."
