Google's AI Chip TPU v5p is 2.8 Times Faster in Training LLMs than its Predecessor


Google promotes its TPUs in the AI hardware market, partnering with small cloud providers to challenge Nvidia's dominance.....
Suning adopts domestic Cambricon AI platform to upgrade its self-developed LingSi model, enhancing tech independence amid global supply chain risks.....
OpenAI recently announced that, despite initial testing, it will not adopt Google's TPU chips on a large scale. TPU (Tensor Processing Unit) is a custom ASIC chip developed by Google for machine learning tasks, designed to accelerate the training and inference of neural networks. TPU uses a dataflow-driven architecture, enabling efficient matrix multiplication pipeline computing and reducing memory access latency. Image source note: The image is AI-generated, provided by the licensing service Midjourney. OpenAI stated that it will continue
Taiwan Semiconductor Manufacturing Company (TSMC), the world's largest semiconductor manufacturer, announced plans to invest at least $100 billion in the United States over the next four years to build multiple new chip fabrication facilities. The announcement was made at a press conference with former US President Trump. TSMC CEO C.C. Wei stated that the investment would primarily focus on expanding its Arizona facilities. Wei added, "We will produce many AI chips to support the progress of Artificial Intelligence." This new commitment builds upon TSMC's previous pledge of a $65 billion US investment.
According to reliable sources, ChatGPT developer OpenAI is advancing its first self-developed artificial intelligence chip project, with the design expected to be completed in the coming months and delivered to TSMC for production. The project plans to use TSMC's advanced 3nm process technology and will be equipped with high bandwidth memory, with an overall architecture similar to NVIDIA's products. This strategic initiative began in October 2023, aimed at addressing the shortage and high costs of AI chips. Sources indicate that OpenAI views this training-focused processor as an enhancement.