Google expands TPU push to challenge Nvidia's hold on AI developers

Traders Union
All News
Business & Wealth
Companies
Google expands TPU to rival Nvidia

Highlights

Google open-sourced TorchTPU, enabling PyTorch applications to run on TPUs with minimal configuration changes and no code rewrites.
Amazon launched TorchNeuron for Trainium chips and contributed to PyTorch's hardware-agnostic development, intensifying competition with Nvidia.
Google introduced a TPU agent suite and updated LiteRT to improve PyTorch portability across its chips, cloud services, and edge devices.

Open-source tools target PyTorch portability

As first reported by TechCrunch, Google said on Tuesday that TorchTPU, its long-awaited version of PyTorch, is now available as an open-source project, allowing developers to run AI applications on the company's Tensor Processing Units, or TPUs.

PyTorch is widely used by AI developers and startups to build, train and run models, and much of its ecosystem has long been optimized for Nvidia GPUs. Google says TorchTPU lets existing PyTorch applications move to TPUs with a simple configuration change, without requiring developers to rewrite their software.

The release comes a day after Amazon introduced TorchNeuron, its own PyTorch variant designed to move applications onto Trainium chips and AWS without rewriting. Amazon also said it is contributing code to the core PyTorch project to separate some features from Nvidia-specific dependencies and make the framework more hardware-agnostic by default.

Google says it has worked with Meta, Hugging Face, Inferact, the maker of vLLM, and RadixArk, the maker of SGLang, to make sure TorchTPU works smoothly with widely used AI development tools.

Cloud competition widens beyond Nvidia hardware

Alongside TorchTPU, Google has launched a TPU agent suite intended to help developers move applications to its chips and cloud services and resolve migration errors during the process. The company has also updated LiteRT, its tool for running AI on phones, laptops and embedded devices, to better support PyTorch-based applications.

The broader push highlights how major cloud groups are trying to weaken Nvidia's default position in AI software development, where many applications were originally built around its hardware. Easier portability does not mean developers will abandon Nvidia at scale immediately, but it does lower switching barriers for companies evaluating cost, performance and cloud options.

In our earlier coverage of the intensifying AI safety debate, we explained how governments and leading developers are weighing stronger safeguards as AI capabilities accelerate. The piece highlighted calls for U.S.-China cooperation on monitoring high-risk uses, proposals for an incident communication channel, and the idea of the UK serving as a neutral venue for international AI safety coordination.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.