Google has introduced two new Tensor Processing Units (TPUs) designed specifically for the demands of agentic AI. One chip is tuned for inference while the other focuses on training, both offering faster performance at a lower cost than previous generations. With these launches, Google positions itself as a serious competitor to Nvidia in the cloud AI hardware market. Although Google Cloud continues to offer Nvidia silicon, the new TPUs mark a strategic push toward a proprietary infrastructure for the agentic era.
Key sources: