Nvidia’s Graphics Processing Hundreds of millions of gamers worldwide are playing with GeForce, a chip that has been powering graphics for nearly three decades. Nvidia’s graphics processing unit has dominated the market, shaping what’s possible in pictures and now even powering ChatGPT, considered revolutionary in AI. Venture capital interest in AI startups has skyrocketed, and the broader world is finally starting to understand the importance of this technology.
Nvidia has invested heavily in AI and is now reaping the rewards, while other chip giants are struggling due to U.S.-China trade tensions and a chip shortage that has weakened demand. However, Nvidia relies on Taiwan Semiconductor Manufacturing Company to manufacture almost all of its chips, which leaves it vulnerable to any potential impact on TSMC caused by U.S.-China relations.
This biggest risk for Nvidia shareholders could keep them up at night. But this isn’t the first time Nvidia has teetered on the leading edge of an uncertain emerging market. Founder and CEO Jensen Huang bet the company on impossible-seeming ventures, leading to near bankruptcy a handful of times. Despite making mistakes, Huang has reinvented the company and invented new technology while competing against larger companies. At Nvidia’s Silicon Valley headquarters, he discusses how he pulled off this latest reinvention and gives a behind-the-scenes look at how Nvidia’s technology powers far more than just gaming.
From Gaming to AI: Nvidia’s Founder and GPU Market Dominance
As a child, 60-year-old Fortune Businessperson of the Year Jensen Huang immigrated from Taiwan to the U.S. and studied engineering at Oregon State and Stanford. Huang met engineers Chris Malachowsky and Curtis Priem at Denny’s in the early 1990s. They dreamed of giving PCs 3D graphics like Jurassic Park.
Huang, Malachowsky, and Priem founded Nvidia in 1993 in a Fremont, California condo. Nvidia named the next version NV after the Latin word for envy, Invidia, hoping to speed up computing so much that everyone would be green with envy.
Nvidia is a rare Silicon Valley giant with its founder still in charge. Huang delivered OpenAI’s first chip inside an AI supercomputer, and his acceleration technology has dominated the graphics processing unit market.
GPUs account for 80% of revenue. These cards accelerate AMD and Intel CPUs by plugging into a PC’s motherboard. They were one of the tens of GPU makers at the time. Because Nvidia worked well with the software community, only they and AMD survived. Not a chip business. End-to-end problem-solving in this business. However, its future was uncertain.
We chose a home run because there weren’t many applications. Video games use computer graphics. Nvidia revolutionized gaming and Hollywood with immediate visual effects rendering. 1997 saw Nvidia’s first high-performance graphics chip. Huang wanted Nvidia to be a fabless chip company, so he outsourced chip production to TSMC.
You’re our hero. Thanks. TSMC’s pioneering work made Nvidia possible. Nvidia released the GeForce 256 in 1999 after laying off most of its employees and nearly going bankrupt.
GPUs generate 80% of Nvidia’s revenue. These GPUs, sold as cards that fit into a computer’s motherboard, boost AMD and Intel CPUs’ computing power. Nvidia, one of many GPU manufacturers at the time, survived due to their strong software community collaboration. Nvidia solves end-to-end problems, not just chips.
Initially, the company’s future was uncertain. Nvidia wisely focused on computer graphics for video games when GPUs had few uses. Nvidia’s immediate visual effects rendering revolutionized Hollywood and gaming. Huang kept Nvidia a fabless chip company by outsourcing the manufacturing of its first high-performance graphics chip to TSMC in 1997.
TSMC and other fabless semiconductor pioneers helped Nvidia succeed. The GeForce 256, Nvidia’s first GPU, was released in 1999 after nearly going bankrupt. Fans and supporters credit Huang and his team for Nvidia’s success.
Nvidia’s focus on AI helps ChatGPT and other things.
Despite some setbacks, Nvidia has 26,000 employees, a new polygon-themed headquarters in Santa Clara, California, and billions of chips for more than graphics. Think data centers, cloud computing, and, most importantly, AI. We’re in every computer company’s cloud. A new application finds you. Over a decade ago, CUDA and GPUs powered AlexNet, AI’s Big Bang. In 2012, a new, accurate neural network won a major image recognition contest.
Parallel processing is ideal for deep learning, where a computer learns without a programmer’s code and lifelike graphics. We wisely supported it company-wide. About a decade ago, we saw that this software approach could change everything, so we changed the company from the bottom up and sideways. All our chips were AI-focused. Six years ago, Bryan Catanzaro was Nvidia’s only deep learning employee.
50 people and growing. For ten years, Wall Street asked Nvidia, Why are you making this investment if no one’s using it? Our market cap valued it at $0. It wasn’t until 2016, ten years after CUDA’s release, that people realized this is a completely different way to write computer programs with transformational speedups that yield breakthrough artificial intelligence results. What’s Nvidia’s AI’s real-world use? Healthcare is big.
Consider faster drug discovery and DNA sequencing. We set a Guinness World Record in genomic sequencing to diagnose these patients and give one trial patient a heart transplant. Thus, a 13-year-old boy who’s thriving and a three-month-old baby with epileptic seizures were prescribed anti-seizure medication. Rafik Anadol’s building-covering Nvidia AI art.
Nvidia’s GPUs were the most sought-after for crypto mining. This isn’t recommended, but crypto mining’s boom-and-bust cycle has caused issues. Gaming cards go out of stock, get bid up, and gaming crashes when the crypto mining boom collapses. Despite Nvidia’s mining GPU, crypto miners bought gaming GPUs, driving prices up.
Last year, Nvidia’s 40-series GPUs were priced much higher than the previous generation, shocking some gamers. Gaming revenue fell 46% in the most recent quarter due to oversupply. Despite the AI boom, Nvidia beat expectations in its most recent earnings report as tech giants like Microsoft and Google fill their data centers with thousands of Nvidia A100s, which train large language models like ChatGPT.
They’re not shipped individually. We send eight groups for nearly $200,000. Nvidia’s DGX A100 server board’s eight Ampere GPUs enable ChatGPT’s lightning-fast and humanlike responses. Thanks to a huge text dataset, I can understand and write about many topics. Competing generative AI companies brag about their Nvidia A100s. Microsoft trained ChatGPT with 10,000.
Their products make adding computing capacity easy. The valley’s currency is computing capacity. Hopper, Ampere’s successor, has shipped. Real-time translation and text-to-image rendering are generative AI applications. This tech creates weird, dangerous deep fake videos, text, and audio. Is Nvidia protecting against some of these bigger fears or building safeguards? Our industry’s AI use precautions are crucial. We’re trying to authenticate content for text, audio, and video.
GPUs that plug into PC motherboards accelerate AMD, and Intel CPUs are Nvidia’s main revenue source. One of many GPU makers, Nvidia survived by working with the software community. Nvidia’s focus on computer graphics and video games revolutionized the industry. Nvidia went public in 1999, but the pandemic drove stock demand.
CUDA, a parallel computing platform released in 2006, helped Nvidia GPUs succeed. In 2020, Nvidia acquired Mellanox for $7 billion despite smartphone market failures. Due to regulatory issues, it abandoned a $40 billion Arm acquisition bid. Apple, Amazon, and many carmakers use Arm CPUs.
Nvidia Faces Market Concerns Amid Export Regulations and Chip Shortage
Despite being at the forefront of the generative AI boom, Nvidia is not immune to market concerns. In October, new regulations were introduced by the U.S. banning the export of leading-edge AI chips to China, which included Nvidia’s A100. Investors have become concerned because approximately a quarter of Nvidia’s revenue comes from mainland China. However, Nvidia’s technology is export-controlled, so compliance with the regulations is necessary. The company had to re-engineer all its products to comply with laws and still be able to serve its commercial customers in China. Nevertheless, Nvidia’s dependence on TSMC in Taiwan is also a major geopolitical risk, with the possibility of China taking over the island and lacking any viable competitor to TSMC. TSMC is currently the only company capable of producing advanced chips in large quantities. However, the U.S. government passed the Chips Act last summer, offering incentives for chip companies to manufacture in the U.S. TSMC is investing $40 billion to build two chip fabrication plants in Arizona. Nvidia will also use the facility for manufacturing.
The chip shortage is another issue that has affected Nvidia, although demand for its AI chips continues to grow due to the chatbot boom. The company’s biggest challenge is staying ahead of competitors, as its customers, like Microsoft, Amazon, Google, Tesla, and Apple, may develop similar products internally. However, according to Jensen, competition is a net positive as the world’s demand for data center power rapidly increases.
Nvidia Focuses on Self-Driving Cars, Robotics, and More.
Nvidia’s focus extends beyond AI and ChatGPT, as evidenced by their use of those technologies in various applications, such as self-driving cars and robots used by Amazon and other companies to optimize their operations. These robots are powered by Nvidia’s Tegra chips, which were once unsuccessful in mobile phones but are now utilized in the world’s largest e-commerce operations. Additionally, Nvidia Drive, used in Tesla Model 3s from 2016 to 2019, is a scalable platform for autonomous driving technology that can be used for everything from simple ADAS to robot axis capable of driving in any condition or weather. Nvidia is also working with other automakers like Mercedes-Benz to develop this technology.
Nvidia has expanded its focus by introducing Grace, a data center CPU, to compete in a new arena. In response to gamers who wish the company would solely focus on their core gaming business, Nvidia’s CEO highlights that their work in physics simulation and artificial intelligence made their recent breakthroughs with GeForce RTX possible. The RTX, released in 2018, features a new technology called ray tracing, which simulates light pathways and everything else with generative AI to take computer graphics and video games to the next level. Ray tracing is currently used in nearly 300 games, including Cyberpunk 2077, Fortnite, Minecraft, and Nvidia’s Geforce GPUs in the cloud, enabling the full-quality streaming of over 1500 games to almost any PC.