.png)
November 21, 2024
Nvidia is experiencing an unprecedented surge. The company has become one of the most valuable corporations in the world in the wake of the AI boom. Jensen Huang, CEO of Nvidia, sees several fundamental trends that continue to fuel this growth.
The first significant trend is the comprehensive modernization of the global IT infrastructure. Huang describes a transformation of enormous scale: The existing, trillion-dollar IT landscape, largely based on CPUs, is currently being upgraded to support machine learning and AI. This change requires a fundamental shift in architecture: away from pure code execution on CPUs, towards the processing of neural networks on GPUs. Huang compares this to a platform shift from Software 1.0 to Software 2.0, where machine learning is used to create AI. This transformation is expected to take several years as companies worldwide need to adapt their data centers.
A second fundamental trend is the emergence of "AI factories" that produce digital intelligence around the clock. Huang sees this as a new industrial revolution that can create a multi-trillion-dollar AI industry. Nvidia's Hopper and Blackwell architectures and platforms like Omniverse play a key role in this development. Demand for Hopper chips is "exceptional" according to Nvidia, and revenue from the H200 chip has more than doubled compared to the previous quarter. Blackwell is in mass production, with demand significantly exceeding supply. Huang attributes the enormous demand to several factors: The increasing number of developers for foundational AI models, the exponentially growing computational needs for pre- and post-training, the increasing number of AI startups, and the success of inference services. Additionally, the introduction of OpenAI's o1 model has produced a new scaling law, called "test-time scaling," which also increases the need for computing power.
In addition to the large cloud providers, a new market for "Sovereign AI" is emerging. Countries and regions are building independent AI infrastructures to meet regional requirements. India, for example, plans to increase the number of Nvidia GPUs tenfold by the end of the year. Japan, with SoftBank, is building one of the most powerful supercomputers based on Nvidia's DGX Blackwell. European countries are also working on regional clouds and AI factories. These developments create new growth markets for Nvidia.
In addition, Nvidia benefits from new optimization techniques such as post-training or test-time scaling, which further increase the demand for computing power. Test-time scaling uses additional resources during runtime to deliver more intelligent responses in real time. OpenAI uses this technique for its new o1 model. Huang emphasizes that these new scaling methods complement the scalability of training foundation models, thus refuting reports that see a limit to scaling possibilities in training.
Bibliographie: https://www.quora.com/What-do-you-think-is-the-significance-of-Nvidia-CEO-Jensen-Huangs-net-worth-increasing-by-3-billion-following-the-jump-in-stock https://www.yahoo.com/tech/nvidia-boss-jensen-huang-predicts-122917467.html https://www.forbes.com/sites/bethkindig/2024/06/07/prediction-nvidia-stock-will-reach-10-trillion-market-cap-by-2030/ https://fortune.com/2024/08/31/nvidia-stock-price-earnings-drivers-outlook-jensen-huang/ https://www.crn.com/news/ai/2024/6-bold-statements-by-nvidia-ceo-jensen-huang-on-ai-s-future https://www.thestreet.com/technology/nvidias-jensen-huang-addressed-three-big-questions-about-ai-future https://www.barrons.com/articles/ai-nvidia-jensen-huang-98f40bd3 https://www.fool.com.au/2024/11/19/3-reasons-to-buy-nvidia-stock-before-november-20-usfeed/ https://www.youtube.com/watch?v=W6Hgi9XBaBU https://www.nytimes.com/2024/02/21/technology/nvidia-earnings.html ```