[SMM Tin Express: Google Plans to Enhance TPU Chip Compatibility with PyTorch to Seize the AI Computing Market]

Published: Dec 18, 2025 09:13

Save

Informed sources revealed that Google (GOOG.O) is advancing a new plan to enhance the performance of its AI chips when running PyTorch, the world's most widely used AI software framework. This move aims to challenge Nvidia's (NVDA.O) long-standing dominance in the AI computing market. Sources indicated that the internal project, codenamed "TorchTPU," is designed to eliminate key barriers hindering the widespread adoption of TPU chips by ensuring full compatibility with PyTorch and providing a more user-friendly development experience for clients who have already built systems using PyTorch. Some sources added that Google is also considering open-sourcing parts of the software to accelerate client adoption. PyTorch, an open-source project strongly backed by Meta Platforms (META.O), is one of the most commonly used tools for developing AI models.

Data Source Statement: Except for publicly available information, all other data are processed by SMM based on publicly available information, market communication, and relying on SMM‘s internal database model. They are for reference only and do not constitute decision-making recommendations.

For any inquiries or to learn more information, please contact: lemonzhao@smm.cn

For more information on how to access our research reports, please contact:service.en@smm.cn

Related News

7 hours ago

Data: SHFE, DCE market movement (Apr 02)

The following table shows the ferrous and nonferrous metals movement on the SHFE and DCE on 02 Apr , 2026

7 hours ago

11 hours ago

Lack of Clear Guidance on the Geopolitical Situation, Repeated Swings in Macro Sentiment Put Futures Under Pressure Again [SMM Tin Midday Commentary]

[SMM Tin Midday Commentary: The Geopolitical Situation Lacks Clear Guidance, and Repeated Swings in Macro Sentiment Put Futures Under Pressure Again]

11 hours ago

12 hours ago

Xidian University Team Unveils Low-Cost SPAD Chip for Short-Wave Infrared Detection

It was learned from Xidian University that the team led by Professor Hu Huiyong at the university successfully developed a single-photon avalanche diode (SPAD) chip based on silicon-germanium technology, significantly reducing the manufacturing cost of short-wave infrared detection technology. This breakthrough is expected to enable high-end chips that originally cost thousands of dollars per unit to enter fields such as smartphones and automotive LiDAR at just one percent of the cost.

12 hours ago