Google Launches TorchTPU to Break Nvidia's Grip on AI Developers

Google Launches TorchTPU to Break Nvidia

Google has launched TorchTPU, an open-source version of the PyTorch AI framework built to run natively on its Tensor Processing Unit chips — a direct challenge to Nvidia's dominance over the software developers use to build AI.

Announced at the PyTorch Conference in San Jose on Tuesday, TorchTPU lets developers move existing AI applications onto Google's TPUs with a simple configuration change instead of rewriting their code. Google built it with help from Meta, Hugging Face, and the makers of popular AI tools vLLM and SGLang. The company also released a TPU agent suite to help developers migrate their apps and fix errors along the way.

The timing is no accident: Amazon launched its own PyTorch port, TorchNeuron, for its Trainium chips just a day earlier. Roughly 85 percent of AI applications are built on PyTorch, and most of its features were designed around Nvidia's GPUs. By making rival chips a near-effortless switch, Google and Amazon are attacking the software moat that has kept developers locked into Nvidia hardware for years.

Source: TechCrunch via MadRobot

Comments

Popular posts from this blog

Fairbairn's Walk-Off Field Goal Seals Wembley Thriller as Texans Edge Jaguars 34-31

Anthropic IPO: Public S-1 Filing Expected Today Ahead of $2 Trillion Listing

Netflix Beats Earnings but Slashes Ad Forecast — Stock Tumbles After Hours