Google Cloud expanding co-engineered infrastructure with NVIDIA

Google Cloud announced an expansion of its co-engineered infrastructure with NVIDIA, including early access to the Vera Rubin platform in the second half of 2026, while also advancing its custom TPU roadmap through a Broadcom partnership secured through 2031, with TPU v7 and v8 generations targeting training and inference workloads.

Google Cloud expanding co-engineered infrastructure with NVIDIA Google Cloud expanding co-engineered infrastructure with NVIDIA including early Vera Rubin access in H2 2026 while pushing custom TPUs v7/v8 generations via Broadcom partnership locked through 2031. TPU focus on training 8t and inference 8i .