Alphabet expands TPU sales to compete with Nvidia

Alphabet is expanding its strategy by offering its Tensor Processing Units (TPUs) directly to external customers. Previously, these specialized AI chips were primarily used internally for Google's own AI workloads and cloud services. This shift means more companies can now purchase and deploy Alphabet's custom AI hardware for their own applications.

This development matters because it introduces a new, significant competitor into the high-demand AI chip market, which has largely been dominated by Nvidia. By making TPUs more accessible, Alphabet aims to capture a larger share of the growing spending on AI infrastructure. This could provide customers with more options beyond current market leaders.

The mechanism involves Alphabet leveraging its existing cloud infrastructure to facilitate the sale and deployment of TPUs. Customers will likely access these TPUs through Google Cloud, integrating them into their existing cloud environments for AI model training and inference. This direct sales model bypasses the traditional reliance on third-party hardware providers for these specific AI accelerators.

This move directly impacts Nvidia (NVDA) by increasing competition in the AI chip sector, potentially affecting its market share and pricing power. It also influences cloud infrastructure spending dynamics, offering Google Cloud (GOOG, GOOGL) a more compelling, vertically integrated AI hardware solution, potentially drawing more enterprise AI workloads to its platform.