What's the driving force behind this impressive expansion?

Simply put, it's the H100.

The H100 is a GPU – the most costly and potent one ever created by Nvidia. It was unveiled last year.

At the grand introduction, Nvidia emphasized that the H100 was designed particularly for AI applications.

This left many people disheartened, questioning, "Why focus on AI with Nvidia chips? Is AI even relevant? AI seems to be at least a decade away."

However, just three months later, ChatGPT was launched.

Consequently, a fierce competition in AI development began, with a multitude of companies striving to create their own AI. GPU demand soared.

Prior to the H100, there were no AI-specific GPUs available on the market.

Consequently, businesses were forced to rely on general-purpose GPUs to develop AI. This process was both time-consuming and required numerous GPUs.

For instance, the training of ChatGPT involved 175 billion data points and the continuous operation of 20,000 Nvidia A100 GPUs for weeks on end.

The Nvida H100 brought about changes, including:

📌 Being 30x faster than the A100, Nvidia's previous top-tier GPU for AI tasks 

📌 Optimization for transformers, which are fundamental to ChatGPT and other LLMs

📌Boasting a substantial 80 billion transistors

The outcome? A skyrocketing demand for Nvidia H100.

From major corporations such as Microsoft and Amazon to innovative startups like OpenAI and Anthropic, everyone is eager to acquire the H100.

As a result, the H100 is the key factor contributing to Nvidia's forecast of a 50% increase in revenues for the upcoming quarter.