When a fan asked Nvidia CEO Jensen Huang to sign her chest earlier this month, that might have been a sign that the hype around the chipmaker may have reached unsustainable heights.
Over the past few years, Nvidia’s computer chips — which have some technical capabilities that make them well suited to AI applications — catapulted the company to new echelons of profitability. Last week, Nvidia briefly became the world’s most valuable company; three days later, it lost that title amid a days-long sell-off of its shares. While its stock price has since recovered somewhat, it’s now the world’s third most valuable company with a market capitalization of $3.1 trillion, after Microsoft and Apple.
The sell-off came amid concern that Nvidia is overvalued. Recently, financial research strategist Jim Reid of Deutsche Bank warned of “signs of over-exuberance” about Nvidia, and Nvidia executives have even sold off some of their holdings in the company.
There are still many reasons to be excited about Nvidia: The company has established itself as an industry-leading chipmaker, reaping the benefits of an early bet on AI that has paid off as chatbots like OpenAI’s ChatGPT have brought broader public attention to the technology.
“It’s very early in the AI race,” said Daniel Newman, CEO of the Futurum Group, a tech research and analysis firm. “But everyone who has been building AI up to this point probably has done at least some of their most important work on Nvidia.”
The stock market has responded accordingly. Nvidia is part of the so-called “Magnificent Seven” tech stocks that accounted for a majority of stock market growth last year. Its stock price had risen nearly 155 percent since January as of the market closing on Wednesday.
But whether Nvidia can continue to replicate that kind of growth depends on advancements in AI, as well as to what extent — and how quickly — businesses will adopt it.
How Nvidia became one of the world’s most important chipmakers
Nvidia has long been considered the premier producer of graphics cards for gaming. However, its graphics processing units (GPUs), the main component of graphics cards, gained popularity amid a rise in cryptocurrency mining, a process that involves solving complex mathematical problems to release new cryptocurrency coins into circulation.
That’s because Nvidia GPUs are highly optimized for what’s called “parallel processing” — basically, dividing up a computationally difficult problem and assigning the various parts to thousands of processor cores on the GPU at once, solving the problem more quickly and efficiently than traditional computing methods.
As it turns out, generative AI also relies on parallel processing. Whenever you query ChatGPT, for example, the AI model has to parse big data sets — the sum total of the world’s text-based online content as of ChatGPT’s last knowledge update — to answer you. To do so in real time and on the scale that companies like OpenAI hope to build out requires parallel processing performed at data centers that house thousands of GPUs.
Nvidia realized what it stood to gain from the GPU needs of generative AI early on. Huang has referred to 2018 as a “bet the company moment” in which Nvidia reimagined the GPU for AI, well before ChatGPT came on the scene. The company structured its research and development and mergers and acquisitions strategies to benefit from a coming AI boom.
“They were playing the game when nobody else was,” Newman said.
In addition to offering GPUs optimized for that purpose, Nvidia created a programming model and parallel computing platform called the Compute Unified Device Architecture (CUDA) that has become the industry standard. This software has made the capabilities of Nvidia GPUs more accessible to developers.
So even as Nvidia’s competitors like AMD and Intel have come to introduce similar offerings, even at lower price points, Nvidia has retained the lion’s share of the GPU market for businesses, in part because developers have gotten used to CUDA and don’t want to switch.
“What [Nvidia] understood very early on is if you want to win in hardware, you got to win in software,” Newman said. “A lot of the developers that are building apps for AI have built them and been comfortable building them using CUDA and running it on Nvidia hardware.”
All of that has positioned Nvidia to capitalize on the ever-growing needs of generative AI.
Can Nvidia keep the good times rolling?
Nvidia’s competitors likely don’t pose any immediate threat to its status as an industry leader.
“In the long run, we expect tech titans to strive to find second sources or in-house solutions to diversify away from Nvidia in AI, but most likely, these efforts will chip away at, but not supplant, Nvidia’s AI dominance,” Brian Colello, a strategist for Morningstar, wrote in a recent report.
However, Nvidia’s ability to sustain the level of growth it has seen in the last year is tied to the future of generative AI and to what extent it can be monetized.
Anyone can currently access ChatGPT for free, though a $20 monthly subscription fee will give you access to its latest and greatest version. But individual subscribers are not currently where the real money is.
Rather, it’s with businesses. And at this point, it’s anyone’s guess how companies will integrate generative AI into their business models in the coming years.
For Nvidia’s growth to be sustainable, major companies like Salesforce or Oracle — which sell software to enterprises — will have to offer new software that will “consume tons of AI” to the point that these large companies are signing annual contracts that give them access to the greatest amount of computing power, Newman said.
“Otherwise, that central thesis of standing up these massive megawatt data centers all over the world full of GPUs becomes a bit of a risk.”
So should you buy Nvidia stock? It depends on how bullish you are about AI and its ability to penetrate the economy.
“We think Nvidia’s prospects will be tied to the AI market, for better or worse, for quite some time,” Collelo writes.
Source: vox.com
