Google Expands TPU Challenge to NVIDIA

Google unveiled TPU variants aimed at regulated workloads and memory-intensive AI, with broader availability reportedly expected in 2027. Coverage reiterates CUDA’s switching-cost advantage and rivals’ potential for gradual share gains, but offers no evidence of near-term NVIDIA demand loss.