Master LLMs with our FREE course in collaboration with Activeloop & Intel Disruptor Initiative. Join now!


“From Gaming to AI: Nvidia’s Pivotal Role in the AI Revolution”
Latest   Machine Learning

“From Gaming to AI: Nvidia’s Pivotal Role in the AI Revolution”

Last Updated on July 15, 2023 by Editorial Team

Author(s): Bhavesh Agone

Originally published on Towards AI.

Nvidia is now worth more than Facebook, Tesla and Netflix. According to Reuters, the stock’s value has tripled in the last eight months. But how did this happen? How did a company that was almost on the verge of bankruptcy in 1997 hit a trillion-dollar market cap in 2023? Let’s find out by understanding what NVIDIA actually does. The processing and calculations of the computer take place on a little chip which is known as the CPU. These include, for instance, the Intel Core i7 or AMD Ryzen processors. The GPU, or graphics processing unit, is the piece of hardware that every gamer, video editor, Crypto miner, and now AI startup is chasing. The GPU takes over and renders sophisticated visuals when a computer is required to do so, such as in video games or 3D modeling. Do you know who came up with the phrase GPU? It was Nvidia, of course. Because of these GPUs, Nvidia’s market cap is now approaching $1 trillion. However, they didn’t get to this achievement overnight.

NVIDIA’s Story

Three engineers, Curtis Preim, Chris Malachowsky, and Jensen Huang, created Nvidia in 1993 because they thought graphics-based computing might become the norm in the future. When NVIDIA almost went bankrupt in 1997 while developing graphics technology for Sega’s game system, that was the company’s first major turning point. ( Like Nintendo, Sega was a company that once competed in the console market). They introduced the Sonic series first, and Nvidia was developing a graphics card for Sega’s 128-bit gaming console Dreamcast. However, after a year of work, Nvidia discovered that its strategy was flawed because Microsoft was set to disclose its own texture mapping strategy, rendering Nvidia’s product incompatible with Windows 95. Nvidia continues to produce this console. Because it wouldn’t sell well and wouldn’t work with Windows, they would eventually go bankrupt. However, if they didn’t build it, they would immediately go bankrupt. NVIDIA CEO Jensen Huang informed Sega’s CEO that NVIDIA would not be able to build the console’s graphics hardware and that Sega should seek out a different partner. He pleaded with the CEO of Sega to give him his full payment since Nvidia was going to collapse without it. Because Sega paid NVIDIA despite the fact that they hadn’t fulfilled their promise to them, it’s practically like asking for money outright. Although Jensen Huang has no idea why they did that, they were able to survive for another six months as a result.

Nvidia began to expand over the following few years. However, because producing these chips was becoming more expensive, Nvidia signed a manufacturing agreement with TSMC. Nvidia aspired to be a fantastic business. Because chip manufacturing in the US is a low-margin industry, they decided against producing their own. TSMC currently produces processors for numerous tech businesses, including Apple, AMD, NVIDIA, and others. Nvidia is significantly reliant on this cooperation, but it has grown difficult. Why?, to know the answer, stick to the blog till the end.

Nvidia: The Powerhouse of Gaming


In the early 2000s, Nvidia obtained a $200 million advance on a deal to produce the graphics hardware for Microsoft’s Xbox. Next year, their revenue reached $1 billion and served as the inspiration for several video games that were released that year. The following turning point was significant for two reasons: first, popular games started to be developed with Nvidia cards in mind, and second, Nvidia’s capabilities extended beyond merely creating graphics cards. They were working along with game designers. The key is that they did more than simply provide graphics cards. They also assisted game creators like Blizzard in optimizing titles on Nvidia technology to ensure smooth frame rates. The video game playing is excellent.

It’s still done today. As a matter of fact, DLSS, one of their upscaling solutions, was recently embraced by a large number of game developers. They used to refer to it as the Way it’s Meant to be Played program. Today, this is a widespread occurrence. Due to the fact that so many games are specifically optimized for Nvidia graphics cards, you often see them powered by Nvidia or a preferred partner, Nvidia at the opening screen of games. GeForce from Nvidia was soon also utilized by PS3. For both Xbox and PlayStation, they were selling graphics equipment. Today, when gamers debate who will win between Xbox and PlayStation, The answer is NVIDIA because their chips are in both of them.

In 2020 most businesses got a sucker punch from Covid 19, but for NVIDIA, it was like they caught the golden snitch. People were suddenly working from home, and when they weren’t working and were still stuck at home, they gamed, and they needed hardware, and PCs became as essential. GPU demand exploded in 2020. Nvidia’s Revenue was just under 11 billion dollars; one year later, the revenue skyrocketed to nearly 27 billion while more than doubling their net income.


This gets us to the next crucial junction where we understand the reason Nvidia is needed today is for A.I. companies. It was cryptocurrency companies in the most recent years. Currently, it’s AI firms. NVIDIA introduced something called CUDA or Compute Unified Device Architecture. No matter what trend emerges, as long as the trend is computing-based, NVIDIA will triumph. It excels in concurrent processing. Typically, GPUs powered by CUDA from Nvidia excel at parallel processing, while CPUs typically excel at serial processing. A GPU has many, many more cores than a CPU, which has up to eight or sixteen cores at most. As a result, while CPUs typically take on a single task, complete it, and then go on to the next task, GPUs have the ability to divide up work. The GPU, for instance, can render certain elements of a scene on various cores when producing a scene(Parallel processing).

Nvidia and AI are a match made in Silicon Heaven; you see, GPUs are the Bedrock of artificial intelligence, and Nvidia just so happens to make some of the best ones in the business. In 2006 Nvidia brought CUDA to the world, a parallel Computing platform and programming model. It lets developers tap into the future potential of GPUs. In layman’s terms, it’s like teaching a racehorse to moonwalk; impressive and incredibly useful for some very specific scenarios. With Cuda, Nvidia turned their GPUs into high-speed Computing powerhouses with the right software; users could now run computations on gigabytes of data with just a few simple queries it took a while for people to catch on, around 10 years.

Nvidia and AI

Nvidia’s GPUs are not just vital for training and operating AI models they are synonymous with them. In ChatGPT, GPUs are needed not only for the initial training of the AI model but every time someone uses chatGPT, so the number of GPUs required will scale with the number of users.

OpenAI probably needed the equivalent of 20,000 DGZ A100 Nvidia GPUs to train the chatGPT model. The further estimate is that OpenAI needs the equivalent of about 30,000 of these systems to support the product’s 100 million active users. As the AI industry grows, Microsoft, Google, Amazon, IBM, and many other companies will have to buy more GPUs.

Nvidia’s Deep Learning Division

Nvidia had already made a name for itself in the gaming industry, but by 2012, Nvidia was doing more than just powering graphics cards. They had new use cases, Datacenters, Cloud GPUs, and of course, training AI models. It turned out that the same parallel processing technique that was beneficial for displaying visuals was great for deep learning, where the computer learns the underlying patterns based on inputs and outputs. Nvidia is aware that this has the potential to fundamentally alter the computing industry. They consequently began concentrating substantially on developing processors designed for AI training. This is what they refer to as their Deep Learning Division.
Do you know what else parallel processing is useful for? Crypto mining!! The crypto-mining process entails solving difficult mathematical puzzles that call for a lot of processing power. Therefore, in the crypto gold rush as well, GPUs with effective parallel processing become essential. People filled up their basements with these cards. Their market capitalization increased as a result in 2021. The AI Gold Rush is experiencing the same thing once more. AI models like GPT and Lama are being trained by businesses like OpenAI, Amazon, and Facebook using entire buildings filled with GPUs. However, Nvidia has been developing unique GPUs, such as the A100, which is geared for deep learning and costs more than $10,000. These are not exactly the same GPUs we use for gaming. ByteDance, the parent company of TikTok, ended up spending a billion dollars as recently as a week ago acquiring chips from NVIDIA. Regardless of who wins the AI race, Nvidia is already winning because most of these companies are using Nvidia’s GPUs to train their model.

Nvidia’s Revenue Model

The major market includes gaming, which is data centers professional visualization, which is 3D modeling software; VFX studios, automotive, and OEMs. Until the fourth quarter of 2021, gaming was their biggest turner. The crypto boom generated a shortage of GPUs, which in turn drove up costs. The gaming sector continued to dominate in 2023 despite a slight decline in gaming revenue and a decline in the number of individuals purchasing gaming GPUs or cards to mine cryptocurrency. However, Nvidia is not very concerned about this, given their data center revenue. It’s growing, All thanks to the AI boom.
Nvidia spends almost $2 billion on research development. it puts most of its effort into research to make the best GPU’s in the world and very little into manufacturing.

Taiwan-China Conflict

Nvidia signed a manufacturing agreement with TSMC. This alliance is becoming somewhat problematic because of the Taiwan-China conflict. Major corporations also want to relocate their chip manufacturing facilities outside of China’s jurisdiction. The CHIPS Act was sponsored by US President Joe Biden and will provide an additional $280 billion in financing to advance semiconductor production and research in the country. In Arizona, TSMC is constructing two chip facilities, the first of which is anticipated to begin operations in 2024. India has also announced a $10 billion manufacturing incentive for semiconductors. If a lot of things work out, India could, over the course of the next ten years, grow into a significant global semiconductor manufacturing hub.


ChatGPT is here and it has changed the way we see the world around us. Many big tech firms are in the race to win the AI Revolution and as I often say, when there are too many big powers involved in any war the technology develops at a great pace. We will soon see many amazing technological advancements from Nvidea in the near future. So guys, buckle up as we’re just scratching the surface of the AI Revolution, and there’s a whole world of possibilities waiting to be explored.

Please feel free to share your thoughts in the comment section. Your suggestions are always welcome.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI

Feedback ↓