In a bold move against Nvidia's dominance, Google has unveiled TorchTPU, its open-source version of PyTorch, allowing AI apps to run on its Tensor Processing Units (TPUs). This announcement follows closely on the heels of Amazon's release of TorchNeuron, designed to port AI apps to its Trainium chip. Both tech giants are making it easier for developers to shift from Nvidia's GPUs, which have long been the default hardware for AI applications.
PyTorch, initially developed by Meta and now under the Linux Foundation, is a favourite among AI developers, powering 85% of AI apps. Historically, PyTorch was tailored to Nvidia's GPUs, but Google and Amazon are shaking things up. Google's TorchTPU offers seamless integration with popular AI development tools, thanks to collaborations with Meta, Hugging Face, and others. Meanwhile, Amazon is contributing code to make PyTorch more hardware-agnostic.
Google's additional tools, like the TPU agent suite and an updated LiteRT, further streamline the transition to its cloud and chips. While these developments don't guarantee an immediate exodus from Nvidia's hardware, they mark a significant push by Google and Amazon to diversify the AI hardware landscape. The battle for AI supremacy is heating up, and the industry is watching closely!
Want to hear more? Join Mal & Matt on the Property AI Report Podcast each week!
Access from your preferred podcast provider by clicking here
Made with TRUST_AI - see the Charter: https://www.modelprop.co.uk/trust-ai