
The podcast discusses the competition in the AI chip market, specifically between Nvidia's GPUs and Google's TPUs. Nvidia's stock fell 3% due to a report suggesting Meta might partner with Google for TPUs. Nvidia responded by asserting its GPUs are a generation ahead and hold over 90% of the market, emphasizing their comprehensive platform, infrastructure, and software. Google's TPUs, though having a potentially better architecture and powering Gemini 3, are ASIC chips optimized for AI training, unlike Nvidia's more general-purpose GPUs. Google currently uses TPUs internally and offers them via Google Cloud, while also remaining a significant customer for Nvidia's GPUs. The discussion concludes with Nvidia's CEO highlighting the "scaling laws" theory, suggesting continued demand for more chips will drive future growth in the AI market.
Sign in to continue reading, translating and more.
Continue