Alphabet’s Google is intensifying efforts to challenge Nvidia’s stronghold on the AI computing market using its specialized AI hardware, Tensor Processing Units (TPUs).
This strategy aims to neutralize Nvidia’s significant advantage derived from its CUDA software, which locks in developers through a powerful ecosystem. Google plans to make TPUs more compatible with PyTorch, a leading open-source AI software framework developed by Meta. This internal initiative is known as "TorchTPU."
By ensuring PyTorch runs smoothly on its TPUs, Google seeks to attract more AI developers and increase hardware adoption. TPU adoption is a critical growth engine for the company's Cloud division. To further accelerate uptake, Google is also considering open-sourcing some of the software components.
The news prompted a market reaction. Following the report, both Alphabet and Nvidia shares traded approximately 3% lower.