Nvidia is an old technology company.
We forget that because it is having a very deserved moment in the sun right now.
Founded in 1993, it reported just 42 employees in 1997 (in the first public filing I could find) and went public in January 1999 with 248 employees.
Here are some of my human takeaways from reading all of their annual filings since their IPO and from some of Jensen’s writings and interviews. I solemnly swear I only used ChatGPT for grammar checking:
Nvidia is constantly reinventing itself. The way they organize the company has changed countless times to address new markets that opened up for GPU computing. The way they describe themselves in the filings year over year is constantly evolving.
They both saw the future and was able to execute on it. The prime example was ImageNet in 2012 when a Nvidia-powered computer was used to run deep learning and beat a human at recognizing images. Plenty of people in Silicon Valley and tech in general knew about AlexNet and ImageNet when it happened, but Nvidia had the ability to extrapolate how machine learning would become the basis for most of the applications in artificial intelligence and turn their GPUs into a parallel computing platform.
They weren’t perfect though. In the early years Nvidia struggled to find product market fit with their first two GPUs and were down to their last attempt before they ran out of money. Fortunately that third generation chip sold, kind of like SpaceX when they were down to that last flight attempt and it worked. Also, they were too optimistic on consumer virtual reality. After we bunch of San Francisco people all got hyped up about VR in 2014 and 2015 post the Oculus winnings, consumer VR didn’t really become a thing. But Nvidia’s Omniverse simulation world for recreating warehouses, city streets, and the entire earth will probably create more real world impact and GDP than all the other consumer VR companies combined. Let’s see.
And then there is Jensen. To say that he was focused as hell and on a lifelong mission to create something special would be a great understatement. Could anyone have picked this out three decades ago or is this survivorship bias at play? Apparently Sequoia and Sutter Hill saw something. Sutter Hill has got to be the highest return over hype ratio venture firm of all time.
Of course it’s impossible to summarize 30 years of development into a blog post so consider this a conversation starter.
Revenues
What jumps out about the chart below?
There was a period from 2008 to 2015 where its revenues did not grow much. 2015 revenues were just 14% higher than 2008 revenues and that was over a 7 year period where other tech companies like Amazon, Facebook, Apple, Google, etc. scaled rapidly.
It also did not grow at all in or even declined in a number of sequential years, as recently as from 2022 to 2023.
Without understanding what’s really going on under the hood at Nvidia, it would have been difficult to stay invested through a number of these periods.
I also typed in their gross profit and operating income into a spreadsheet like any good little banking analyst would do but you can get those financials in a lot of other places. It is, however, worth noting that from 2023 to 2024 their revenues more than doubled, growing at 126% which is just insane at this scale and their gross margins increased from an already very good 57% to 73% year over year. Have we ever seen such a banner year for a public company in the history of capitalism?
Often attributed to Bill Gate is the quote, "We always overestimate the change that will occur in the next two years and underestimate the change that will occur in the next ten"
It would be strange to achieve artificial general intelligence in the next two years strictly speaking, but what if we are still underestimating its impact in the next ten years?
Videos & Blogs
2011 Stanford Online: Jen-Hsun Huang: Stanford student and Entrepreneur, co-founder and CEO of NVIDIA
Accelerating AI with GPUs: A New Computing Model January 12, 2016
By 2011, AI researchers around the world had discovered NVIDIA GPUs.
As it turned out, 12 NVIDIA GPUs could deliver the deep-learning performance of 2,000 CPUs.
Baidu, Google, Facebook, Microsoft were the first adopters of NVIDIA GPUs for deep learning.
In just two years, the number of companies NVIDIA collaborates with on deep learning has jumped nearly 35x to over 3,400 companies.
Popular Science recently called the GPU “the workhorse of modern A.I.” We agree.
The Intelligent Industrial Revolution October 24, 2016
Some 1,500 AI startups have cropped up. This explosive growth has fueled demand for GTCs all over the world.
In my three decades in the computer industry, none has held more potential, or been more fun. The era of AI has begun.
In short, we offer an end-to-end AI computing platform — from GPU to deep learning software and algorithms, from training systems to in-car AI computers, from cloud to data center to PC to robots.
Letter From Jensen: Building the New NVIDIA Together April 30, 2020
I am thrilled to announce that today, April 27, 2020, NVIDIA and Mellanox are officially one company!
NVIDIA and Mellanox are a perfect combination.
AI, the most powerful technology force of our time, is a fundamentally new computing model that is reshaping data center architectures.
The fastest computing and the fastest networking go hand in hand.
We are thrilled to become one family and invent the future together.
NVIDIA to Build Earth-2 Supercomputer to See Our Future November 12, 2021
The earth is warming.
To make our future a reality today, simulation is the answer.
We can achieve million-x speedups by combining three technologies: GPU-accelerated computing; deep learning and breakthroughs in physics-informed neural networks; and AI supercomputers, along with vast quantities of observed and model data to learn from.
2023 Nvidia: An Overnight Success Story 30 Years in the Making
(Roelof Botha) The crucible moments in today’s episode center on Nvidia’s willingness to bet on unproven markets years ahead of time.
(Jensen Huang) I mean, the list of mistakes that we made, and that I made, in the first three years of the company, you could really write a book.
(Jensen Huang) I said let’s work backwards. If we get one shot, what do we have to do to make sure that that one shot was perfect?
(Andrew Ng) So around 2008, 2009, my students and I started to work on and push the idea that GPUs could be used for deep learning, for neural networks.
(Roelof Botha) Nvidia made a crucible decision that would change not only its own trajectory but that of the entire technology industry. They would commit to AI computing.
(Jensen Huang) Well, if you can solve it in about thirty years, you’re probably going to solve it in five. That’s the big realization. That’s groundbreaking.
2024 Acquired Podcast: NVIDIA CEO Jensen Huang
Reports & Filings: start from the bottom to read chronologically up and look at the picture of Nvidia’s first “office”.
NVIDIA is now a full-stack computing infrastructure company with data-center-scale offerings that are reshaping industry.
In 2012, the AlexNet neural network, trained on NVIDIA GPUs, won the ImageNet computer image recognition competition, marking the “Big Bang” moment of AI.
With a strong engineering culture, we drive fast, yet harmonized, product and technology innovations in all dimensions of computing including silicon, systems, networking, software and algorithms.
More than half of our engineers work on software.
While we work to enhance the resiliency and redundancy of our supply chain, which is currently concentrated in the Asia-Pacific region, new and existing export controls or changes to existing export controls could limit alternative manufacturing locations and negatively impact our business.
We want NVIDIA to be a place where people can build their careers over their lifetime. Our employees tend to come and stay. In fiscal year 2024, our overall turnover rate was 2.7%.
Under our agreement with Microsoft for the Xbox, if someone makes an offer to purchase at least 30% of our outstanding common stock, Microsoft may have first and last rights of refusal to purchase the stock.
Generative AI is a new computing platform like the PC, internet, mobile, and cloud.
The result of over four years of partnership and collaboration with TSMC, ASML, and Synopsys, cuLitho accelerates computational lithography by over 40X and paves the way for the industry to go to 2nm and beyond.
Generative AI has created a sense of urgency in companies everywhere to reimagine their products and business models.
ChatGPT has taken the world by storm.
More than 40,000 companies are already running on NVIDIA to speed up, scale up, and unlock previously impossible-to-solve problems.
There are 3.8 million developers worldwide using CUDA and our other software tools to help deploy our technology in our target markets.
AI is the most powerful technology force the world has ever known.
Grace is our entry into the CPU market.
Starting this year, we are offering enterprise software license and support models for NVIDIA AI, NVIDIA DRIVE, and NVIDIA Omniverse.
We built NVIDIA like a computing stack, or neural network, in four layers: hardware, system software, platform software, and applications.
Including GPUs and networking, NVIDIA powers over 70%, and 8 of the top 10, supercomputers on the global TOP500 list.
On February 8, 2022, NVIDIA and SoftBank announced the termination of the Share Purchase Agreement whereby NVIDIA would have acquired Arm from SoftBank.
We plan to build Earth-2, an AI supercomputer dedicated to predicting the impacts of climate change.
We want to build a one-of-a-kind company that invents the future, builds amazing technologies, and strives to achieve the highest level of craft.
Researchers at UC San Diego did the largest, longest, and most accurate simulation of a coronavirus.
We may have advanced science a decade in the past year.
Developers have downloaded CUDA and NVIDIA AI software over 24 million times.
RTX is a home run—our most successful GPU generation ever.
In September, we announced an agreement to acquire Arm for $40 billion to create the world's premier computing company.
The cornerstone of NVIDIA AI is the re-invention of the GPU to become Tensor Core GPUs.
We reinvented every layer of computing for AI.
Starting in 2024, the entire Mercedes-Benz fleet will be powered by NVIDIA DRIVE
So, we must work to stay relevant, fight the gravity of goodenough, and seek the extraordinary.
With our acquisition of Mellanox in fiscal year 2021, we strengthened our end-to-end expertise in data center architectures, positioning us for a future when the data center is the new unit of computing.
NVIDIA specializes in accelerated computing, solving important challenges beyond the reach of normal computers.
NVIDIA Omniverse is a new collaboration platform to unite these project pipelines and let designers share assets seamlessly across creative applications.
NVIDIA-powered supercomputers all over the world are being called into service to search for new drug compounds that can treat or vaccinate against the virus.
NVIDIA RTX is a home run. We have created the best reason to upgrade graphics since the invention of the GPU.
We made a giant bet, reimagined the computing stack, and invented a whole new kind of computer.
I can’t imagine more important work.
NVIDIA now powers the fastest supercomputers in the United States, Europe, and Japan.
Eventually, every machine that moves will be autonomous—including hundreds of millions of warehouse forklifts, farming machines, and delivery robots.
We believe the advent of AV will soon revolutionize the transportation industry.
For them, we’ve built the equivalent of a time machine.
We expect AI to spur a wave of social progress unmatched since the industrial revolution.
NVIDIA has a platform strategy, bringing together hardware, system software, programmable algorithms, libraries, systems, and services to create unique value for the markets we serve.
A decade ago, we set out to transform the GPU into a powerful computing platform—a specialized tool for the da Vincis and Einsteins of our time.
Every industry has awoken to AI.
We leverage our powerful platform by focusing on the hardest problems that only we can solve.
NVIDIA is a learning machine. For over two decades, we’ve continuously reinvented ourselves.
This is undoubtedly the most exciting time in our industry.
We, at NVIDIA, are seizing the moment.
We provide a complete, end-to-end GPU computing platform for deep learning, addressing both training and inferencing.
We aspire to build one of the world’s great companies, respected and admired for the amazing work we do and the impact we make.
VR is coming next.
Since the dawn of the computer age, artificial intelligence has been the final frontier.
In 2015, GPU-accelerated deep learning systems achieved superhuman levels of perception for the first time.
Our culture is our greatest achievement.
Our invention of CUDA has enabled the GPU’s parallel processing capabilities to be harnessed to accelerate general purpose computing.
Our work opens up new universes to explore, enables amazing creativity and discovery, and powers what were once science fiction inventions like self-learning machines and self-driving cars.
NVIDIA powers 8 million cars on the road today, a figure that will reach 25 million in five years.
NVIDIA GPUs have been broadly adopted in deep learning, a branch of the longstanding field of artificial intelligence.
I am delighted by NVIDIA’s transformation this year.
Our business model has three elements: creating NVIDIA-branded products and services, offering our processors to original equipment manufacturers, or OEMs, and licensing our intellectual property.
At its core is the GPU, which we invented 15 years ago and serves as the visual cortex of modern computers.
We returned $1.07 billion to shareholders through a combination of share repurchases and cash dividends.
Our lead was borne out of the invention and creativity of the world’s most talented engineers, who joined NVIDIA to be challenged to do their best work.
Our two reporting segments - GPU and Tegra Processor - leverage a single, unified architecture.
We are seeing the “computerization” of all electronic devices around us.
CUDA 5, has been downloaded once a minute, on average, since its launch last year.
We recently announced Project SHIELD, the first Android device designed for gaming.
By extending visual computing into the cloud, NVIDIA is expanding the application of the GPU beyond the PC into the server and datacenter.
We are one of the very few companies from the PC industry to have made the leap to the mobile industry.
The computing world is experiencing its biggest transformation in decades.
Of the world’s Top500 supercomputers, 35 are now powered by NVIDIA GPUs—triple last year’s number.
NVIDIA solutions are based on two important technologies: the GPU and the mobile processor.
Mobile processors incorporate central processing unit, or CPU, and GPU technologies to deliver an entire computer system on a single chip, or system-on-chip.
Fans have had the GeForce logo tattooed on their arm and shaved in their hair.
NVIDIA’s latest reinvention is based on energy efficient high-performance processors that are 100-times more energy efficient than a PC.
We are strategically investing in three major areas—visual computing, high performance computing and mobile computing.
NVIDIA Corporation helped awaken the world to the power of computer graphics when it invented the graphics processor unit, or GPU, in 1999.
Because of limited 40nm wafer foundry capacity, plus supplier challenges related to 40nm process manufacturing yields, we have been forced to allocate our available 40nm product supply among our customers. We are currently working with our foundry partners to address these challenges.
We believe we are in an era of GPU computing, where our CUDA parallel processing architecture can accelerate compute-intensive applications by significant multiples over that of a central processing unit, or CPU, alone.
There are currently over 25,000 developers around the world using CUDA.
We generated 87%, 89% and 86% of our revenue for fiscal years 2009, 2008 and 2007, respectively, from sales to customers outside the United States and other Americas.
In fiscal year 2008, we also introduced NVIDIA Tesla, our entry into the high-performance computing industry.
We also announced Compute Unified Device Architecture, or CUDA, a new mode of operation on GPUs where the computational power of the GPU can be utilized for computation-intensive applications.
Our objective is to be the leading supplier of performance GPUs, MCPs and handheld GPUs and application processors.
GoForce handheld GPUs implement innovative design techniques, both inside the chips and at the system level, which result in high performance and long battery life.
Our objective is to be one of the most important and influential technology companies in the world.
Our WMP product line is primarily incorporated into multimedia-rich cellular phones.
We invest significant resources in the development of relationships with industry leaders, including Cadence Design Systems, Inc., and Synopsys, Inc., often assisting these companies in the product definition of their new products.
WMPs: wireless media processors
We have three major product groups: GPUs, MCPs and UMPs.
In addition, we sell a two-chip platform processor for use in Microsoft’s Xbox video game console.
Our MCP product family, known as nForce, supports desktop PCs, notebook PCs and professional workstations.
Our UMP product family, known as GoForce, supports handheld personal digital assistants, or PDAs, and cellular phones.
On August 14, 2003, Microsoft announced that it had entered into an agreement with one of our competitors to develop technology for future Xbox products and services.
On February 24, 2004, we announced that we will be one of the first semiconductor companies to manufacture select up-coming GPUs at TSMC's 0.11 um (micron) process technology.
GPUs: graphics processing units
MCPs: media and communications processors
UMPs: ultra-low power media processors
We are one of the world’s largest “fabless” semiconductor companies, supplying graphics and media communications processors and related software that are integral to personal computers, or PCs, professional workstations and digital entertainment platforms.
Our mission is to be the most important visual computing company in the world.
However, on March 26, 2003, we announced that we have formed a multi-year strategic alliance under which IBM will manufacture our next-generation GeForce GPUs.
We have five major product brands: GeForce, nForce, GeForce Go, Quadro and TNT2.
The GeForce3 is designed for the ultimate PC enthusiast and is the first member of our third-generation GeForce family.
Fabricated on a .15 micron process technology, the 57 million transistor GeForce3 delivers 33 million sustained triangles per second.
Our GPUs, the GeForce 256 and NVIDIA Quadro, process over 200 billion operations per second and increase the PC's ability to render high-definition 3D scenes in real-time.
The agreement provides that in April 2000, Microsoft will pay us $200 million as an advance against graphics chip purchases and for licensing our technology.
The visually engaging and interactive nature of 3D graphics responds to consumers' demands for a convincing simulation of reality beyond what is possible with traditional 2D graphics.
We have a "fabless" manufacturing strategy whereby we employ world class suppliers for all phases of the manufacturing process, including fabrication, assembly and testing.
As of January 31, 1999, we had 248 employees, 117 of whom were engaged in engineering and 131 of whom were engaged in sales, marketing, operations and administrative positions.
NVIDIA designed the RIVA128 graphics processor to enable PC OEMs and add-in board manufacturers to build award-winning products by delivering state-of-the- art interactive 3D graphics capability to end users while maintaining affordable prices.
As of December 31, 1997, the Company had 92 employees as compared to 42 employees as of December 31, 1996, and the Company expects that the number of its employees will increase substantially over the next 12 months.
Additional paid-in capital $22,902 (12/31/1997)
Via Wikipedia: The Denny's roadside diner in East San Jose, c. 2023, where Nvidia's three co-founders agreed to start the company in 1993