The Nvidia advantage

Watching the below video, I realised that it is computationally cheaper to generate an answer to a question, than to send the same query to a search engine... this means LLMs enabled by Nvidia will replace search engines like Google. And now, Nvidia released NIMs. You can watch the video below to understand what Nvidia NIMs are. Essentially, the layers of software that Nvidia built over 20 years looks like this: 1) CUDA - low level library to distribute a computing task in parallell, and thereby speeding it up with factors of 1000x. 2) High level libraries built on top of CUDA that researchers already know, used to do common tasks such as fluid simulations, image recognition, speech recognition, etc. 3) And now NIMs, which enables anyone to run pre-trained generative models with all dependencies etc on your own hardware, or in the Nvidia network of AI factories - essentially making Nvidia an AI-cloud provider. NIMs are, in short, Nvidia's AppStore moment. https://www.youtube.com/watch?v=IurALhiB6Ko