Dear Shareholders, NVIDIA had an excellent year. Full-year revenue was a record $26.91 billion, up 61 percent from $16.68 billion a year ago. Gross margins expanded to 64.9 percent, and earnings per share were $3.85, up 123 percent from a year ago. We returned $399 million during the year to shareholders through quarterly cash dividends.
Our top priority remains the continued expansion of NVIDIA accelerated computing—creating software, systems, and chips to serve new markets. NVIDIA grows through any combination of these. New chips open new markets. For example, the new Orin robotics processor chip is the central computer for a new generation of logistics robots, delivery robots, and self-driving cars, trucks, and taxis. And Grace is a new CPU architecture designed for ultra-fast data processing, the essential operation of AI and machine learning. Grace is our entry into the CPU market.
New software opens new markets. NVIDIA has sped up computing a million times over the past decade by GPU-accelerating algorithms, optimizing across the full stack, and scaling across the entire data center. NVIDIA SDKs, or software development kits, capture the complex computer science and engineering into expertly crafted CUDA libraries. Our cuQuantum SDK running on DGX computers is the ideal simulation system for quantum computing, chemistry, and cybersecurity researchers. And while quantum computing may still be a few decades away, NVIDIA cuOpt is a real-time combinatorial complexity planner that operations, logistics, and delivery services can use today.
NVIDIA SDKs are the heart and soul of accelerated computing. New SDKs connect us to new opportunities. NVIDIA has hundreds of SDKs that form several core platforms: NVIDIA RTX for graphics, NVIDIA HPC for scientific computing, NVIDIA AI for data science and AI, NVIDIA DRIVE for autonomous vehicles, and NVIDIA Omniverse, our physically accurate simulation platform for developing and operating automated fleets of cars and trucks, and automating places like warehouses and factories.
Starting this year, we are offering enterprise software license and support models for NVIDIA AI, NVIDIA DRIVE, and NVIDIA Omniverse.
New software, new systems, and new chips can create markets. Machine learning algorithms developed on TensorFlow and PyTorch, and processed by NVIDIA AI running on our DGX systems with Tensor Core GPUs, revolutionized AI. Omniverse running on OVX systems with NVIDIA Spectrum networking chips and systems will be added to NVIDIA's platform to open the next wave of AI.
DEAR NVIDIANS AND STAKEHOLDERS
We’ve expanded our platform in every dimension and direction—more developers, more applications, and new customers in new industries. NVIDIA now serves over three million developers and is available from every major computer maker and cloud, and has been adopted by over 25,000 companies in 100 trillion dollars' worth of industries.
We built NVIDIA like a computing stack, or neural network, in four layers: hardware, system software, platform software, and applications. Each layer is open to computer makers, service providers, and developers to integrate with their offerings.
We innovated across every layer this year. At our virtual developers conference GTC 2022, we announced new GPU, CPU, and networking chips and systems—our largest rollout to date.
The Hopper-architecture-based NVIDIA H100 GPU has arrived, and its performance is unbelievable. It's the new engine of the world's AI infrastructure. At four petaflops of AI power, H100 performance is an order of magnitude beyond that of our NVIDIA Ampere-based A100. Designed to support the next wave of AI adoption, it introduces a set of breakthrough security features like GPU isolation and confidential computing, dynamic programming to speed complex science problems, direct memory connections, and high-speed networking.
We also introduced the Grace CPU Superchip—our entry into CPUs. With an unbelievable one-terabyte-per-second memory bandwidth, Grace moves and processes mountains of data. It's ideal for AI infrastructure, cloud, and scientific computing.
In addition to GPU and CPU news, we announced NVIDIA Spectrum-4—the world's first 800 Gbps networking switch. With Spectrum for Ethernet, Quantum for InfiniBand, NVLink for multi-node DGX, and our DOCA networking, storage, and security infrastructure software stack, NVIDIA is ready to build out the world's AI infrastructure, from end to end.
NVIDIA GPU, CPU, and networking systems are some of the industry's most advanced technologies and form the foundation of our computing platform.
We now have more than 450 NVIDIA AI libraries and software development kits to serve industries such as gaming, design, quantum computing, AI, 5G/6G, and robotics.
It was a breakthrough year for NVIDIA accelerated computing.
NVIDIA RTX REINVENTS COMPUTER GRAPHICS, BLURRING REAL AND PHYSICAL WORLDS
Computer graphics is the driving force of NVIDIA. Through the decades, we've repeatedly reinvented computer graphics—the GPU, programmable shading, material simulation, and, in 2018, RTX real-time ray tracing and AI image synthesis.
The arrival of NVIDIA RTX real-time ray tracing, approximating the physics of light and materials to generate photorealistic virtual worlds, accelerated the industry to warp speed and pulled us all 10 years into the future.
Today, RTX is everywhere and reinventing every application of 3D graphics. More than 250 games and apps use RTX, from Fortnite to Adobe Photoshop. Architecture walkthroughs, product design, product configurators, and virtual showrooms are now photorealistic. NVIDIA RTX made real-time photorealistic 3D graphics possible. The line between computer graphics and reality, and therefore the digital and physical worlds, will become increasingly blurred.
THE AI INFRASTRUCTURE OF AI COMPANIES
Fast computers running AI algorithms can learn patterns and relationships from data to write software that was previously impossible. Knowledge of this new way of developing software and products is spreading quickly. Companies from nearly every industry are processing and refining their raw data to make AI software. They are manufacturers of intelligence, and data centers are their factories.
We work with over 25,000 companies, applying AI to reinvent their industries—from consumer internet companies to those in healthcare, manufacturing, logistics, retail, financial services, energy, design, and transportation. AI is the most impactful technology force of our time. We are all hands on deck to advance the NVIDIA AI platform, solve new problems, and make AI more accessible—from PC, cloud, enterprise data centers, the industrial edge, robotic systems, and scientific supercomputers.
THE NEXT WAVE OF AI
This year, bringing Omniverse to market was a huge achievement and opened a new chapter for NVIDIA. Omniverse is a 3D simulation engine to build and connect virtual worlds. In Omniverse, designers can collaborate with other designers, and even AIs and robots, to create virtual worlds. Digital twins of robots can hone their skills in virtual worlds before performing tasks in the physical world.
An automobile manufacturer like BMW can build and test a new factory with a digital twin before laying the first brick and configuring the first assembly line. Once the actual factory is operating, logistics planners will use the digital twin to prototype new process flows and software. That software can be deployed into the physical factory when optimizations demonstrate improved efficiency. Digital twins will be essential to designing and operating future software-defined plants.
Large industrial companies have already created fully operational digital twins in Omniverse. Siemens Energy is developing a digital twin for predictive maintenance at power plants that could help save utility providers an estimated $1.7 billion a year. Ericsson is simulating the interplay between 5G cells and the environment. PepsiCo is using digital twins to improve the efficiency and sustainability of its more than 600 distribution centers. Amazon is using digital twins to optimize the design and flow of its warehouses and train more intelligent robotic solutions.
The Omniverse ecosystem has expanded quickly in its first year. More than 80 software tools connect to Omniverse, across design and rendering, robotics, digital twin, and industrial automation ecosystems. Omniverse has been downloaded 150,000 times by individual creators and evaluated by over 700 companies. And now, Omniverse Cloud adds a one-click capability to reach tens of millions of designers, creators, roboticists, and AI researchers.
Nearly 30 years of NVIDIA graphics, physics, simulation, AI, and computing expertise made Omniverse possible. We now have the engine for the next wave of AI.
ROBOTIC SYSTEMS—AI AUTOMATES THE WORLD
Today's AI centers on perception and pattern recognition. AI can recognize an image, understand speech, suggest a song, or recommend an item to buy. A new wave of AI is dawning in which robots perceive the environment, reason about where they are, where they need to go and what they need to do, and then develop a plan to do it.
Robots will take many forms, physical and digital. Some will stock shelves in warehouses or build cars. Others will be medical instruments that integrate real-time sensor data and assist with medical procedures. Self-driving cars are robotic vehicles with digital robots inside, riding along with us, helping with driving and navigation tasks.
NVIDIA Isaac is our platform for industrial robots. More than 700 customers and partners use Isaac to develop robots for e-commerce fulfillment, retail automation, cleaning, and last-mile delivery.
NVIDIA Clara Holoscan is for AI-powered medical systems—an open, programmable platform built to medical-grade specification for next-generation robotic imaging and surgical instruments.
NVIDIA DRIVE is for autonomous vehicles—a full-stack open platform for AV development. The automotive industry is enormous, with 100 million cars sold each year and an installed base of over a billion vehicles. With an $11 billion design win pipeline over the next six years, Auto has several upcoming inflection points that we believe will create the next multibillion-dollar business for NVIDIA.
The NVIDIA Orin robotics computer powers Isaac, Holoscan, and DRIVE. Orin is in production and already powering 25 electric vehicle makers building autonomous vehicles.
The NVIDIA DRIVE AV computer and software will power Mercedes-Benz's new fleet in 2024. And beginning in 2025, all new Jaguar and Land Rover vehicles will be powered by NVIDIA DRIVE AV.
THE POWER OF A MILLION-X TO BATTLE CLIMATE CHANGE
We are living in a transformational time in the field of computer science. Our invention of the GPU and CUDA democratized scientific computing. NVIDIA AI computing scaled across a data center can solve scientific challenges a million-x more in speed and scale, opening the door to tackle some of the world's grandest challenges, like drug discovery and climate science.
We announced last year our Earth-2 initiative—a digital twin of our planet. Earth-2 will be a simulation system to predict the complex multi-physics of Earth's atmosphere, land, sea, and ice caps at sufficiently high resolution to predict regional impacts of human actions over decades.
Earth-2 will require the invention of new chips, supercomputers, and AI algorithms that can predict climate physics millions to billions times faster than today.
Earth-2's implications for science and society can be massive. And there is early evidence we can succeed. Researchers at NVIDIA and universities Berkeley, Caltech, Michigan, Purdue, and Rice have developed a weather forecasting AI model called FourCastNet that can predict weather events such as hurricanes, atmospheric rivers, and extreme rain up to 45,000x faster. What previously took a classical numerical simulation a year to complete now takes just minutes.
Collaborating with climate scientists worldwide, we hope to create a system that lets researchers simulate climate change while providing a platform to predict the impact of mitigation and adaptation strategies.
Earth-2 will pull together the brightest minds to tackle one of humanity's greatest threats.
LOOKING AHEAD
We had a fantastic year, but nothing makes me prouder than the incredible people who have made NVIDIA the company it is today. We want our company to be where they can do their life's work, so it's a true honor that we were ranked No. 1 on Glassdoor's Best Places to Work list for large U.S. companies. Employees alone drive this ranking, and I am so grateful.
Our employees support the communities in which they work and live. They joined the company in contributing more than $22 million to charitable causes in fiscal year 2022.
Together, we drove advances this year in AI, HPC, gaming, creative design, autonomous vehicles, and robotics—some of the world's most impactful areas.
We enter the upcoming year with strong momentum. Demand for our products is exceptional. We made substantial strides in diversifying and expanding our supply base. Our execution has never been better. And we go into this year with the largest wave of new products in our history.
I want to thank NVIDIA developers, partners, customers, and families for the amazing work you do. Exciting new frontiers lie ahead. Let's seek them out together.
Jensen Huang
CEO and Founder, NVIDIA
April 2022
Read the original letter here.
