Ever wondered why NVIDIA’s Keynote speeches are packed with attendees?

NVIDIA’s annual conferences are like concerts, for real. When I first realized it, I thought “Wait, wait, wait…what’s going on here.” It was so hard to wrap my head around because I had never seen any business conferences having the level of attendees that NVIDIA has. But anyone, even a 9-year-old, can understand why things are the way they are by just asking themselves the following simple question, which I did to do so that helped me come to the right conclusion: “What actually NVIDIA is and what does it do?”

The reason NVIDIA’s Keynote speeches have tens of thousands of attendees is that NVIDIA attracts a wide array of customers—from AI companies to the agricultural sector to electric vehicle brands to pharmaceutical companies…you name it. So it’s obvious that when NVIDIA launches its new products and technology, these different companies and institutions want to attend the conferences so they can be the early adopters of the new technology.

Last year in 2024, Jensen Huang, the co-founder and CEO of NVIDIA, delivered an electrifying speech at NVIDIA’s GPU Technology Conference (GTC) in front of 11,000 attendees, to which NVIDIA Research Manager Jim Fan called Jensen Huang the new Taylor Swift.

I don’t think any other tech company gets this level of attendees to their annual conferences. However, interestingly, NVIDIA definitely does, and does every single year. But it’s only possible because NVIDIA is nothing short of a platform for platforms. NVIDIA is a platform that is designed to help other platforms win by providing exceptional products and services—at scale.

Now this goes back to that simple question: “What actually NVIDIA is and what does it do?” The simple answer is: NVIDIA is a tech company that designs and manufactures chips, software, and systems for countless industries and companies—AI, gaming, blockchain, healthcare, agriculture, data centers, the automotive industry along with probably 100 other industries.

Said differently, NVIDIA’s customer is everyone who comes from different industries. For example, the company serves its products and services to OpenAI, Microsoft, Google, Amazon, Tesla, Waymo, Toyota, Dell, HP, Siemens, Deepmind, MasterCard, Paypal, Sony, Nintendo, AT&T, Walmart, Alibaba along with hundreds of thousands of other companies worldwide.

NVIDIA truly is a platform for platforms.

But it wasn’t too long ago when NVIDIA became a platform for platforms and took the internet by storm. This is also why most people assume that NVIDIA is an overnight success company, which to a degree is true, but it’s not the whole truth. NVIDIA is a three-decade-old company, though the company didn’t see significant growth in the first two decades of its founding. It was only a few years ago when NVIDIA really started outcompeting other tech companies and joined the Trillion Dollar Club. The company wasn’t even a trillion-dollar company until 2023, so it’s not too long ago, which also means the majority of (~80%) the company’s valuation came within 2 years, which is pretty remarkable and mind-boggling.

So is this an overnight success for NVIDIA? Definitely NOT. The company has literally been grinding for decades until it finally saw success a few years ago. It also won’t be wrong to say that the company had already seen the future two decades ago that we’re seeing right now, and has been preparing itself to build world-class infrastructures and products so that it could take advantage of today’s massive opportunity.

What NVIDIA has accomplished is nothing short of extraordinary.

So this week’s deep dive is dedicated to NVIDIA. I’m going to take you through the company’s founding story, how it was preparing itself for the future, and how it grew in three main phases, which will prove the idea of NVIDIA being an overnight success company wrong. Surprisingly, It’s the opposite, NVIDIA is one of those companies that teaches you about thinking long-term, having patience, and doing things differently.

Get your popcorn ready and let’s dive in!

Growth Phase 1

Let’s start with NVIDIA’s founding story.

NVIDIA was founded on April 5, 1993, by Jensen Huang, a Taiwanese-American electrical engineer who was previously director of CoreWare at LSI and designer at AMD, Chris Malachowsky, an engineer who worked at Sun Microsystem, and Curtis Priem, who was previously a senior staff engineer and graphic chip designer at IBM and Sun Microsystem.

The three guys decided to start the company at Danny’s roadside diner on Berryessa Road in East San Jose. NVIDIA initially had no name as they called their files NV, which meant “Next Version.” At one point, Malachowsky and Priem wanted to call the company NVision, but the name was already taken by a toilet paper company. Later, Jensen suggested the name Nvidia, from Invidia, the Latin word for “Envy” and headquartered the company in Sunnyvale, California.

The company's goal was to redefine the way computers process visuals, especially 3D animations. It was the early 1990s, and the co-founders being engineers themselves having technical backgrounds knew that most PCs relied on CPUs to handle graphical tasks, which was highly inefficient at that time—believing that there was potential to disrupt the CPU industry.

The company secured $20 million in total funding from Sutter Hill Ventures and Sequoia Capital and spent its first two years developing its first big product, NV1, which was released in 1995, hoping to bring 3D graphics to the gaming and multimedia market. But unfortunately, the product commercially flopped. However, it taught co-founders valuable lessons, which helped them build the next breakthrough product.

NVIDIA introduced RIVA 128, their second product, which became a megahit—selling a million units in four months. But what’s interesting is, by the time RIVA 128 was launched, the company had only one month of payroll for employees, which meant if RIVA 128 had failed, the company would have gone out of business. This “Special” moment brought life to the company and For years, Jensen Huang would open staff presentations with the following line that unofficially became the company motto: “Our company is thirty days from going out of business.”

The company was set to launch its next-generation product.

NVIDIA introduced its next product GeForce 256 in 1999, which not only skyrocketed the company’s growth but also changed the gaming industry forever. GeForce 256 was the first Graphic Processing Unit (GPU) which had a dedicated transformer and lightning engine, significantly improving the 3D graphics performance. This innovation of NVIDIA set the new standard for how 3D graphics should work and what modern GPUs should be capable of.

Because of this breakthrough moment and successful product launch, NVIDIA won the contract to develop the graphics hardware for Microsoft’s Xbox game console, which earned NVIDIA a $200 million advance, the company acquired its competitor 3dfx, and went public the same year.

Now one thing to notice is that, it was very early for the company, though NVIDIA was growing, but it wasn’t “A Platform for platforms.” Yet, the company initially, solely focused on a niche market, which was the gaming industry. But things get interesting as NVIDIA scales and taps into new markets again, and again, and again.

Growth Phase 2

NVIDIA’s Growth Phase 1 was just the tip of the iceberg.

NVIDIA’s most transformative innovation came 13 years after founding the company in 2006 with the introduction of CUDA (Compute Unified Device Architecture), a parallel computing platform and programming model that transformed GPUs into general-purpose computing engines. CUDA enabled developers to leverage the massively parallel GPU tasks beyond graphics—opening up possibilities in scientific research, engineering simulations, financial modeling, and eventually machine learning.

While CPUs had long been the backbone of general-purpose computing, CUDA showed that GPUs could outperform CPUs by orders of magnitude in highly parallel workloads. However, developing CUDA wasn’t a smooth ride for NVIDIA. The company had to build an ecosystem around CUDA, including developer tools, libraries, and extensive training programs. Which, NVIDIA did and it paid very well as new companies started using the platform.

If you’re still confused about what the heck CUDA is, here’s a simple explanation: CUDA is a platform built for developers to program using programming languages like C, C++, Fortran, Python, and MATLAB. The platform offers 150 CUDA-based libraries, SDKs, and profiling and optimization tools. The beautification of the platform is that anyone from any domain can use the platform to get their things done. For example, pharmaceutical companies use CUDA to discover new promising treatments, car companies use it to speed up autonomous driving, and online stores and brands use CUDA to analyze customers' purchases and buyer data to make recommendations and place ads.

This platform single-handedly opened up an entirely new market for NVIDIA—attracting a wide array of customers from multiple domains—from finance to agriculture to healthcare to geography to cryptocurrency to data science…the list goes on and on. The best and unique feature of CUDA is its parallel programming model. Before CUDA, using GPUs for non-graphics tasks required "tricking" the GPU by making your computation look like a graphics problem. CUDA provided a straightforward way to write programs that could harness the GPU's massively parallel processing power.

Do you know OpenAI’s ChatGPT is built on CUDA?

Exactly, you heard it right. ChatGPT is built on AI models that are trained and run on NVIDIA GPUs using CUDA. Since OpenAI’s large language models such as GPT-4 or GPT-4o require massive computational resources, they rely on NVIDIA’s high-performance GPU, which is highly optimized for AI training and inference.

Another example: Waymo, a Google car company heavily relies on NVIDIA’s GPUs and CUDA for AI training and simulations to use the combination of high-definition mapping, LiDAR, and deep learning models, which require massive computational power. Waymo trains its AI models on NVIDIA’s GPUs using CUDA and also uses NVIDIA hardware in its self-driving fleet.

And as you can imagine this innovation again skyrocketed the company’s growth reaching hundreds of thousands of new customers and businesses from different domains. All because CUDA was built for “Any” business, institutions, enterprises, and organizations to build, manage, and innovate their technologies.

Now this was the time when NVIDIA really became the platform for platforms because businesses and organizations could use NVIDIA products and services along with CUDA to build, run, manage, and scale their products and technologies.

And with that, Forbes named NVIDIA Company of the Year for 2007.

Growth Phase 3

This is the growth phase that usually starts around 2012.

This is the phase that drove the highest growth for the company. Because from the early 2010s to the early 2020s, not one, but NVIDIA became the go-to service provider in different industries. We “Now” know that this is the era of AI and Cryptocurrency, but do you know NVIDIA has been deeply preparing and building infrastructure for decades to leverage today’s opportunity? All companies around the globe, doesn’t really matter which industry they operate in, most of the time, for most businesses and organizations NVIDIA happens to be the go-to service provider for chips, GPUs, operations works, and data centers, especially if they want the facilities of deep learning and parallel processing computing through CUDA.

Here’s what happened in Growth Phase 3:

#1: AI: The era of AI started forming in 2013, just around when NVIDIA was harnessing its infrastructure and GPUs. With years of work, and all thanks to NVIDIA’s powerful GPUs, today almost every major AI breakthrough is backed by NVIDIA’s GPUs from deep learning to natural language processing and autonomous vehicles. NVIDIA’s Tensor Core architecture has made AI models more efficient, reducing the training time, and increasing inference speeds. NVIDIA’s AI ecosystem, including software frameworks like TensorRT and hardware solutions like DGX systems, endures a competitive edge in enterprise AI adoption.

#2: Mining: Blockchain technology and Cryptocurrency were pretty new back in 2013. But today? Bitcoin has already hit $100,000, an all-time high. Last year Stripe acquired a StableCoin platform Bridge for $1.1 billion to help people and businesses make transactions using Blockchain technology. And there is no going back from here. But NVIDIA has been taking advantage of Blockchain and Cryptocurrency since 2013. Crypto transactions work on blockchain, which requires mining, and that’s where NVIDIA’s chips (GPUs) comes into the game because they are the best for mining as it requires a ton of power and efficiency. The more popular and adaptive blockchain and cryptocurrency will become, the more NVIDIA will benefit from it. Though it is benefiting from it, when the crypto boom really started in 2018, 2021, and then recently in 2024, the demand for NVIDIA’s GPUs skyrocketed. And not only that, developers and businesses that wanted to build blockchain software and technologies, guess where they went to? Exactly. NVIDIA’s CUDA platform.

#3: Metaverse: Ever wondered how Apple Vision Pro was developed? It was built through NVIDIA’s Omniverse. This is another project NVIDIA started in 2022 to tap into the Metaverse, which they call the next stage of the internet and how people will interact with humans and objects in the future. Though, the concept of Metaverse is fairly new and people are still skeptical about it. But on the other hand, seeing new technologies like Apple Vision Pro and Meta’s Orion proved that Metaverse is going to be a real thing as these tech giants are pouring money into it. Why Omniverse? Only Omniverse provides a collaboration and simulation platform for 3D content creation, leveraging NVIDIA’s strengths in AI, physics, simulation, and real-time rendering.

#4: Humanoid Robots: You may only know about Tesla’s Optimus, but do you know NVIDIA is also developing and building its own infrastructure and systems to build humanoid robots? And I guess they are not far behind in the game. The company has already built a powerful platform, Isaac Platform, and developed NVIDIA Cosmos to make humanoid robots possible. And I can’t tell you, I mean how many, hundreds of thousands of companies worldwide using Isaac and Cosmos platform to build different types of humanoid robots and robotic machines.

#5: Data Centers: Another great initiative NVIDIA made in recent years is building its own data centers. NVIDIA heavily focused on building its own data centers and storage facilities to streamline not just its revenue potential but also to become self-reliable. Today its powerful A100 and H100 GPUs are integral to AI, cloud computing, and HPC (High-performance computing) workloads, which makes NVIDIA an indispensable partner for cloud giants and research institutes.

This video gives a good summary about what NVIDIA is building:

So all this to say, what are the results?

Well, most of NVIDIA’s valuation, revenue, and profit came in the Growth Phase 3. The company’s total market cap went from $140 billion in 2019 to $3+ trillion in 2024, whereas its total revenue jumped from $10.9 billion in 2019 to $113 billion in 2024. That means more than 80% of the company’s market valuation came after 2020 and in the last two years or so the company has 10x its total revenue, which I think is pretty astonishing.

Let’s be honest, NVIDIA couldn’t be more successful. NVIDIA says that more than 4 million developers now create tens of thousands of applications for accelerated computing, more than 40,000 companies use NVIDIA AI technologies, with 15,000 global startups in NVIDIA Inception.

However, any business, organization, or institution can build what NVIDIA has already built—powerful GPUs, platforms, and models. But most aren’t going to build, and would rather prefer using NVIDIA’s products and services because building what NVIDIA has already built would not only cost a ton of money but also time and energy, and maybe failures too. And that’s why they have to rely on NVIDIA, which makes it the platform for platforms.

Thanks for reading, catch you on the next one.