Alphabet’s Google and Meta Platforms are collaborating on "TorchTPU," an initiative designed to weaken Nvidia’s stronghold on the AI software market. The project aims to improve the compatibility of Google’s Tensor Processing Units (TPUs) with PyTorch. PyTorch, an open-source AI framework developed by Meta, has historically been associated with Nvidia’s CUDA platform.
By enhancing compatibility, Google seeks to position its TPUs as a more viable alternative to Nvidia’s market-leading GPUs for enterprise customers and developers. Analysts cite Nvidia's powerful software moat as its primary competitive advantage in the booming AI industry. This collaboration represents a significant attempt by Google to erode that advantage.
The news triggered a negative market reaction for both tech giants involved in the AI chip competition. Following the report, shares of both Google and Nvidia fell by approximately 3% in trading.