AivexaNewsSearch
AI news for builders and product teamsChecked every hour

Google has joined Amazon in attacking Nvidia’s stronghold with AI app developers

Collected Oct 20, 2026

Google announced on Tuesday that TorchTPU, its version of PyTorch for Google's Tensor Processing Units (TPUs), is now available as an open-source project. The release positions Google alongside Amazon in attempting to loosen Nvidia's grip on the AI industry, where most AI applications are written by default to run on Nvidia GPUs, according to TechCrunch.

PyTorch, originally developed by Meta in 2016 and now managed by the Linux Foundation, is a widely used open-source toolkit for building, training, and running AI models. Some estimates put its adoption at 85% of AI apps. Because Nvidia's GPUs were already the standard for AI research when PyTorch was created, many of its features were built specifically for Nvidia hardware.

TorchTPU allows existing PyTorch applications to run on Google's TPUs and, by extension, on Google Cloud. Developers can migrate without rewriting their apps, using a simple configuration change. This is a notable shift: porting between hardware backends has traditionally required code changes to handle different memory layouts, kernels, and libraries. By offering a drop-in path, Google lowers the barrier for teams that want to try TPUs or reduce reliance on Nvidia.

The timing is significant. On Monday, Amazon released its own PyTorch variant, TorchNeuron, which ports AI apps to its Trainium chip and AWS without rewriting. Amazon also said it is submitting code to the main PyTorch project to decouple various features from Nvidia chips, making the framework more hardware-agnostic by default. That upstream contribution could benefit all hardware vendors, including Google, by making PyTorch less Nvidia-centric.

Google may have an edge in ecosystem integration. It says it worked directly with Meta, Hugging Face, Inferact (maker of vLLM), RadixArk (maker of SGLang), and others to ensure TorchTPU's features and code libraries work well with popular AI development tools, just as the original PyTorch does. This suggests Google is trying to match the compatibility that developers expect from PyTorch on Nvidia.

Beyond TorchTPU, Google launched a TPU agent suite to help developers move their apps to Google's chips and cloud and fix errors along the way. It also updated LiteRT, its tool for running AI on phones, laptops, and embedded devices, to work better with PyTorch apps. These moves indicate Google is targeting both cloud and edge deployments.

In practice, this lowers the switching cost for PyTorch users. A developer who wants to experiment with TPUs can now do so with a configuration change, potentially avoiding weeks of porting work. However, performance and feature parity may still vary; the source does not claim that TorchTPU matches Nvidia's performance or supports every PyTorch feature. A likely trade-off is that early adopters may encounter gaps in operator coverage or optimization, which the TPU agent suite presumably aims to address. AWS's TorchNeuron offers a similar value proposition for Trainium, so developers now have two credible alternatives to consider.

None of this means AI app developers will automatically ditch Nvidia hardware en masse just because alternatives are easier to reach. Nvidia's ecosystem, including CUDA and years of optimized libraries, remains deeply entrenched. But Amazon and Google are clearly working to erode that advantage. The upstream code contribution from Amazon could also make PyTorch more portable overall, which would benefit the entire hardware ecosystem.

Why it matters: Developers and product teams building on PyTorch now have a simpler path to run their AI workloads on Google TPUs or AWS Trainium, potentially reducing cloud costs or avoiding vendor lock-in. This is inference: teams that have been Nvidia-only may reassess their infrastructure choices, especially for inference at scale where price-performance matters. The competition may also pressure Nvidia to improve its software stack or pricing, though the source does not speculate on such outcomes.

Read at TechCrunch · AI

Based on reporting from the original publisher. Visit the source for full context and later updates.

Publisher excerpt

Google has joined Amazon in releasing a developer tool that makes it easy for PyTorch users to move their apps off Nvidia GPUs.