started · updated
Google and Amazon launch tools to challenge Nvidia's AI dominance
Google and Amazon have both released new open-source tools designed to allow AI developers to run PyTorch workloads on their respective proprietary chips without requiring extensive code rewrites. This move directly challenges Nvidia's market leadership, as most AI applications are currently optimized for Nvidia's GPUs and CUDA software ecosystem.
Google announced TorchTPU, which enables the seamless migration of existing PyTorch applications to Google's Tensor Processing Units (TPUs) and Google Cloud via a simple configuration change. Google has collaborated with Meta, Hugging Face, Inferact, and RadixArk to ensure compatibility with major AI development tools.
Amazon Web Services (AWS) introduced TorchNeuron, allowing developers to port PyTorch workloads to Amazon's Trainium processors, including the Trainium2 and Trainium3 models. Amazon is also contributing code to the main PyTorch project to decouple specific features from Nvidia hardware, aiming to make the framework more hardware-agnostic by default.
PyTorch, originally developed by Meta and now managed by the Linux Foundation, is used by an estimated 85% of AI applications. By lowering the engineering barrier to switch hardware, both tech giants aim to reduce the industry's heavy reliance on Nvidia's ecosystem.
Entities
AWS · Amazon · Annapurna Labs · Google · Meta · Nvidia · PyTorch · Trainium
Claims
What the coverage asserts, and how many sources carry each claim.
- [● 5 SOURCES] TorchTPU allows running PyTorch workloads on Google TPUs with a simple configuration change. www.forux.it · www.it-boltwise.de · www.ithome.com · www.newsbytesapp.com · techcrunch.com
- [● 10 SOURCES] Amazon launched TorchNeuron as an open-source toolkit. www.prodiris.fr · technews.tw · insidetelecom.com · primanews.org · borncity.com · +5 more
- [● 5 SOURCES] Google released TorchTPU as an open-source project. www.forux.it · www.it-boltwise.de · www.ithome.com · www.newsbytesapp.com · techcrunch.com
- [● 10 SOURCES] TorchNeuron enables running PyTorch workloads on AWS Trainium chips without major code rewrites. www.prodiris.fr · technews.tw · insidetelecom.com · primanews.org · borncity.com · +5 more
- [● 10 SOURCES] PyTorch is used by approximately 85% of AI applications. www.forux.it · insidetelecom.com · primanews.org · borncity.com · www.ithome.com · +5 more
- [● 3 SOURCES] Google collaborated with Meta, Hugging Face, Inferact, and RadixArk for TorchTPU compatibility. www.ithome.com · www.newsbytesapp.com · techcrunch.com
- [● 5 SOURCES] TorchNeuron supports Trainium2 and Trainium3 processors. www.prodiris.fr · technews.tw · borncity.com · siliconangle.com · thenextweb.com
- [● 3 SOURCES] Amazon is submitting code to the main PyTorch project to decouple features from Nvidia chips. www.forux.it · technews.tw · techcrunch.com