Huang Renxun became good friends with everyone in Silicon Valley this year, becoming the only person to visit the three conference stages of Microsoft, Amazon, and Google. Why? Because everyone needs GPUs to power their AI dreams.
Graphics processing units were once just coveted by PC gamers looking to maximize their home setups, but now they’re the crown jewels of the world’s most profitable tech companies. It’s a unique location, but how did Huang get here? What does he do with this once-in-a-lifetime opportunity when he’s not on the conference stage?
Jealousy of the Valley
Nvidia dominated the AI chip market this year: By 2023, more than 80% of the chips used to train and run AI models will be made by Nvidia, according to Baird analysts. When you hear about big tech companies like Apple spending “millions of dollars a day“Training artificial intelligence, a lot of that goes to Nvidia, and if more equipment is available, more money will flow in.” Jen-Hsun Huang named his company Nvidia, after the Latin word for “jealousy,” and now Everyone in Silicon Valley must be envious of his company.
Microsoft and Meta Acquisition 150,000 Nvidia H100 and H800 units This year, the cost is about $30,000 each. Google, Amazon and Oracle bought as many of them as they could, about 50,000 each.Joe Biden deliberately set us foreign policy Limit Nvidia chips from falling into the hands of Chinese companies. 2023 is the year that powerful people know Jen-Hsun Huang’s name.
Big tech executives lined up on stage to shake Jansen’s hand and smile for photos, but there was a tense atmosphere in the air. Soon after Huang walked off the stage in his signature black leather jacket, his biggest customer quickly announced its own competition chip to train artificial intelligence models. Big Tech doesn’t want to be influenced by anyone. Being the biggest star in Silicon Valley might mean you’re friends with everyone now, but it also gives you a huge purpose.
GPU History
Jen-Hsun Huang’s central role in the artificial intelligence revolution began early. The Nvidia CEO is worth an estimated $43 billion, but his favorite place to hold meetings is a booth at Denny’s. More than a decade ago, he made a bet at one of the booths that GPUs, not CPUs, were the future of gaming.
A GPU can perform multiple tasks simultaneously, while a CPU can only perform one task at a time. 2009 MythBusters Demo Still the best, comparing the GPU to 1,100 simultaneously fired paint guns to recreate the Mona Lisa in less than a second. This is how the GPU works, performing parallel tasks, while the CPU is more like a paint gun trying to paint a smiley face, creating something that is less impressive and takes more time.
GPUs revolutionized the gaming industry, making video game graphics look far better than anything made with a CPU. It turns out that, through Jensen’s incredible luck or foresight, GPUs would prove crucial to training artificial intelligence, which, like graphics rendering, requires billions of simultaneous tasks to build large neural networks. GPUs have become an essential part of any company doing artificial intelligence, and every other chipmaker is years behind Nvidia.
Competitor or customer?
This year, everyone is making headlines for their forays into artificial intelligence chip production.Microsoft has Maya 100Google has TPU v5pAmazon has its Training 2Meta announced its mia. They all sound great, but the reality is that none of them can compare to Nvidia’s GPUs. Nvidia has a years-long lead over its competitors, and all of these companies will be training most of their AI models on Huang’s chips for the foreseeable future.
“The vertical integration initiatives we’re seeing are very, very low-key,” Tristan Gerra, senior research analyst for semiconductors at Baird, told Gizmodo. “Meta won’t have its own chips until 2025. AWS has discussed developing its own GPUs, which we know is very complex and challenging and will take several years.”
Gerra said we will see internal efforts to develop AI chips in the coming years, but they will be very limited. Nvidia may only lose a few percentage points of market share next year.
“For Nvidia, there’s really no comparison to the high double-digit growth that we’re going to see from them and the entire field over the next few years,” Guerra said.
In the long term, Nvidia may have to worry about big tech companies as chip competitors, but for now, they’re just customers. Big tech companies can market their artificial intelligence chips at conferences as much as they want, and Jen-Hsun Huang won’t lose any sleep. By the time Nvidia caught up, they had bigger plans for computing domination.
Jensen Huang’s vision of turning leverage into an empire
Nvidia is selling its GPUs to cloud providers such as Google, Microsoft and AWS in exchange for hosting Nvidia’s own AI cloud services. DGX Cloud. So just as these companies are chasing Nvidia’s artificial intelligence chips, Huang is developing cloud products to replace them. But who will win?
Big Tech’s best friend, and possibly worst enemy, told the New York Times at the DealBook Summit that artificial intelligence has revolutionized computing. Huang hopes to use his position to dominate a new era of computing.
“We’re at the beginning of a new generation of computing. It hasn’t been reinvented in 60 years, which is why it’s so important,” Huang said. Currently, he points out, computing is mostly about retrieval—you just ask your phone to retrieve files from a server somewhere. He said the future of computing will involve retrieval and generation driven by artificial intelligence.
The 60-year computing revolution he was referring to was led by Intel, the chip company that perfected the mass-market CPU chip founded by Gordon Moore and Bob Noyce in 1958. Intel has greatly improved human computing power by placing many small transistors on silicon computer chips. Jensen Huang believes Nvidia and GPUs are at the center of the next revolution; the next Intel.
“You can’t solve this new way of computing just by designing a chip,” Huang said. “From the network to the switching, to the way computers are designed, to the chip itself, all the software that sits on top of it, and the way it’s put together. This is a big deal because it’s a complete reinvention of the computer industry. “
The shift Jensen is talking about is consistent with the future of technology described by Sam Altman. During OpenAI’s keynote, Altman described a future where humans ask computers to complete tasks for you, rather than completing tasks on the computer. They describe AI generation as not just a feature, but a core operating system for computers moving forward.
Huang counts himself among the first to recognize this new future, which involves new data centers, new computer designs and new coding languages to power them. Jensen Huang and Nvidia are betting on this vision and want to be the technology company building the future of computing. In 5 to 10 years, the big tech companies may catch up to where GPUs are today, but that’s just the beginning for Jen-Hsun Huang.
“It’s hard for people to understand it,” Huang said. “But that’s the great observation we made. It contains a chip but has nothing to do with that chip.”
