What a fantastic day to remember! The whole Wall Street was on edge predicting the dooms day scenario and a drop in the stock price up to 11% and dragged the stock and the price down for almost two days bringing it from the mighty top of $740 down to $674 before the close.
Yet, when the humble NVIDIA team announced the results afterhours, all the doubts perished in the sand and pop goes the stock, up by 9% to $735 after hours on Feb 21, 2024 and to $762 (up 13%) pre-market on Feb 22.
In this post, we would like to first congratulate the CEO Jensen Huang and the NVIDIA team for an incredible quarter and summarize some key takeaways from the call for the onset of the 4th Industrial revolution - AI charged and led by NVIDIA.
Some Key Takeaways
We are at the beginning of a new Computing era, accelerating the pace of adoption from General Purpose Computing (GPC) to Accelerated Computing (AC), thereby driving improved performance, TCO and Energy Efficiency. These improvements will accelerate new applications, software and companies built on Generative AI. The current existing recommender systems powering the internet based on CPU's will need to be upgraded to GPU accelerated computing.
A New generation of data centers, called the AI factories are going to disrupt the current $1T Install Datacenters and is expected to grow 2X in the next 5 years.
In addition to the Compute and Store, the AI Generation factories, AI Data factories powered by NVIDIA will be the key catalysts that will convert the massive scale of Industrial and Commercial Internet raw data into Digital Intelligence and valuable tokens.
A diverse and a thriving ecosystem is evolving that enables Gen AI Inferencing, GPU specialized CSP's, AI powered Enterprise Software Platforms, AI powered Consumer Internet Companies and Industrial Gen AI.
There will be AI Generative Factories for every industry and every country/region.
Sovereign AI will be the next evolution replicated from the success of AI in the US across different countries and regions that can customize the LLM's and foundation models to regional knowledge, language, culture and history. As AI factories evolve and scale, the regions will own, protect, transform and provision their data and digital intelligence.
While the current AI today is powered and built on GPC by Hyperscalers, it will transform into Industry specific AI factories powered by NVIDIA AI Enterprise or simply NVDIA OS for AI.
NVIDIA OS will maintain, optimize, manage, patch, tune and containerize the software by Industry that is required across Public Cloud, Private Cloud, Hybrid Cloud and On-Prem.
NVIDIA data centers will be powered by Hopper Architecture and Infiniban Networking that uses SpectrumX (1.6X increase in networking efficiency for AI)
The H200 servers are performing 2X compared to H100 on the Inferencing capacity.
NVIDIA DGX cloud will work seamlessly with Amazon AWS, Google Cloud, Microsoft Azure and will host NVIDIA AI R&D, Custom Model development.
Google Gemma, the open source LLM version of Gemini is powered through the powerful NVIDIA Inferencing Engine.
Close to 80 Automotive companies are using NVIDIA drive infrastructure solution to ingest, curate, label, train and simulate autonomous driving.
NVIDIA Clara and NVIDIA BioNemo are powering the foundation models enabling computer aided drug discovery and making recursion availability.
The Hopper Transfer Engine is expected to bring amazing breakthroughs in LLM's as they begin to use the CUDA architecture.
We believe that these developments will continue to accelerate the pace of AI revolution and adoption across industries and regions in the very near future.
As we are monitoring these trends, if you need any assistance in transforming your organization and be AI ready, reach us out at contact@katalyststreet.com