top of page
  • Writer's pictureKatie Richman

AI is Powered by GPUs. It’s Time to Understand What They Are.


If AI is your overbooked restaurant, GPUs are your new expanded pizza kitchen.
If AI is your overbooked restaurant, GPUs are your new expanded pizza kitchen.

If AI is your overbooked restaurant, GPUs are your new expanded pizza kitchen.

Since starting at NVIDIA about a month ago, I’ve spent a lot of time schooling myself on GPUs — especially when it comes to why they are crucial to AI advancement … and thus, why NVIDIA’s stock is so astronomical right now. GPUs. I knew it had to do with processing graphics for gaming. I knew it was somehow powering the metaverse under the hood. But it took me starting a job at NVIDIA to buckle down and really understand how GPUs have anything to do with generative AI and large language models.


To understand why GPUs are essential to generative AI (and all machine learning), we need to talk about how computers work. Before you yawn, you should know that this metaphor was generated through a brainstorm with ChatGPT. I asked for some analogies that will help us truly get it…and ChatGPT suggested a pizza kitchen. Want to understand why AI is Powered by GPUs? Read on.


1. Your computer is like a restaurant.


Your restaurant has 2 kitchens: the general kitchen and the pizza kitchen.
Your restaurant has 2 kitchens: the general kitchen and the pizza kitchen.

Your restaurant has 2 kitchens: the general kitchen and the pizza kitchen.

Imagine your computer is your own restaurant, and business is booming. You have lots of different things on the menu, including delicious custom pizzas. The problem? Making pizzas is super labor-intensive for your chefs, who have to stop what they were doing to make one. That’s why you launched a pizza kitchen.


Think of your computer like a restaurant that employs two key kitchens:1) the CPU (Central Processing Unit) kitchen and the GPU (Graphics Processing Unit) kitchen.

The CPU is like the main kitchen where general food preparation happens, and it handles a variety of tasks. The GPU, on the other hand, is our specialized pizza kitchen, designed for specific tasks like making pizzas quickly and efficiently.


2. The CPU Kitchen: Jack of All Trades


You can’t constantly derail your head chef (your computer’s) with custom time-intensive pizza orders. Let’s hire some pizza chefs (GPUs).
You can’t constantly derail your head chef (your computer’s) with custom time-intensive pizza orders. Let’s hire some pizza chefs (GPUs).

You can’t constantly derail your head chef (your computer’s) with custom time-intensive pizza orders. Let’s hire some pizza chefs (GPUs).


The CPU kitchen is versatile, equipped to handle a range of tasks, from chopping vegetables to simmering sauces. It’s like a skilled chef who can cook everything but might take more time to prepare for a large banquet … or to make lots of pizzas.

In computer terms, the CPU handles general computing tasks, from running your operating system to browsing the web.


3. The GPU Pizza Kitchen: The Specialist

Then the restaurant built the GPU pizza kitchen. Here, the focus is on making pizzas (processing graphics and complex computations). Like a team of chefs each handling part of the pizza-making process, the GPU is designed for parallel processing. It’s great at doing many specialized tasks simultaneously, like kneading multiple dough balls or adding toppings to several pizzas at once.


4. Parallel Processing: Many Hands Make Light Work

Parallel processing in the GPU is like having multiple chefs working on different pizzas at the same time. Each chef (GPU core) focuses on one part of the task, such as rolling dough, slicing toppings or baking.

This teamwork leads to a lot of pizzas (data) being prepared quickly, which is essential for tasks like playing video games or running AI algorithms.


5. Why AI is Powered by GPUs

In the world of AI, data is the key ingredient. Preparing this data requires specialized handling — akin to making a lot of pizzas quickly in our GPU kitchen. GPUs can process large chunks of data simultaneously, making them ideal for training AI models, where vast amounts of data need to be processed efficiently.


6. The Rise of NVIDIA: A Top Chef in the GPU Kitchen

NVIDIA has emerged as a top chef in the GPU kitchen. NVIDIA’s expertise in making high-quality, efficient GPUs has made them crucial in the AI and gaming industries. Having NVIDIA’s GPUs is like hiring a team of the best pizza chefs, ensuring that complex data tasks are handled swiftly and efficiently … and don’t bog down the main kitchen from preparing everything else.


7. The Combo of the CPU and the GPU

In a well-run restaurant, the main kitchen and the pizza kitchen work in harmony. Similarly, in a computer, the CPU and GPU complement each other. The CPU handles general tasks, while the GPU takes on more specialized, data-intensive tasks.

Together, they make your computer fast, powerful, and versatile.

Just as a pizza kitchen is essential for a restaurant specializing in Italian cuisine, GPUs are vital for computers handling graphics-intensive tasks (running video games) and complex computations (AI operations). The GPU pizza kitchen is required to run AI computations…or your entire restaurant will fail.

0 views0 comments
bottom of page