When a fan asked Nvidia CEO Jensen Huang to sign his chest earlier this month, it may have been a sign that the hype around the chipmaker had reached unsustainable heights.
Over the past few years, NVIDIA's computer chips have evolved technical capabilities that make them ideal for AI applications, propelling the company to new levels of profitability. Last week, NVIDIA briefly became the world's most valuable company, before losing that title three days later after a days-long sell-off in its stock. The company's shares have since recovered somewhat, but with a market capitalization of $3.1 trillion, it is now the world's third-largest company after Microsoft and Apple.
The selling comes amid concerns that Nvidia is overvalued. Deutsche Bank financial research strategist Jim Reed recently warned of “signs of excessive exuberance” in the company, leading Nvidia executives to sell off some of their shares.
There's plenty of reason to be excited about Nvidia: The company has established itself as a leading chipmaker and has benefited from early investments in AI, with chatbots like OpenAI's ChatGPT bringing wider public attention to the technology.
“The AI race is just getting started,” said Daniel Newman, CEO of technology research and analysis firm Futurum Group, “but everyone who's ever built AI has done at least some of their most important work at Nvidia.”
The stock market responded accordingly. Nvidia is one of the so-called “Magnificent Seven” tech stocks that accounted for much of the stock market's growth last year. As of the market's close on Wednesday, the company's shares had risen nearly 155% since January.
But whether Nvidia can replicate that kind of growth going forward will depend on advances in AI and how much and how quickly companies adopt it.
How Nvidia became the world's leading chipmaker
Nvidia has long been considered the leading manufacturer of gaming graphics cards, but the key component of the company's graphics cards – GPUs (graphics processing units) – has gained popularity amid the rise of cryptocurrency mining, the process of solving complex mathematical problems to generate new cryptocurrency coins for circulation.
That's because Nvidia GPUs are highly optimized for something called “parallel processing,” which basically involves breaking down a computationally difficult problem and assigning different parts to the GPU's thousands of processor cores at once, solving the problem more quickly and efficiently than traditional computational methods.
In fact, generative AI also relies on parallel processing. For example, when you query ChatGPT, the AI model has to parse a big data set (the total amount of text-based online content in the world at the time of ChatGPT's latest knowledge update) to get the answer. To do this in real time and at the scale that companies like OpenAI want to build, data centers with thousands of GPUs need to run the parallel processing.
Nvidia recognized early on the benefits to be gained from generative AI's GPU needs. Huang calls 2018 the “moment we bet the company on,” when Nvidia reimagined GPUs for AI, long before ChatGPT came along. The company built its R&D and mergers and acquisitions strategy to benefit from the coming AI boom.
“They were playing when no one else was,” Newman said.
In addition to providing purpose-optimized GPUs, Nvidia also developed an industry-standard programming model and parallel computing platform called Compute Unified Device Architecture (CUDA), software that makes it easier for developers to harness the power of Nvidia GPUs.
So even as Nvidia's competitors such as AMD and Intel have introduced similar products at lower prices, Nvidia has maintained the lion's share of the enterprise GPU market, in part because developers have grown accustomed to CUDA and are unwilling to switch.
“what [Nvidia] “It was understood very early on that if you want to win in hardware, you have to win in software too,” Newman says. “Most developers who are building apps for AI are happy to develop them using CUDA and run them on Nvidia hardware.”
All of this has put Nvidia in a position to capitalize on the growing need for generative AI.
Can Nvidia continue its strong run?
Nvidia's competitors don't appear to pose an immediate threat to the company's position as an industry leader.
“Long term, we expect tech giants to try to move away from Nvidia in AI and find second-source or in-house solutions, but these efforts will likely reduce, not replace, Nvidia's AI advantage,” Morningstar strategist Brian Colello wrote in a recent report.
But whether Nvidia can maintain the same levels of growth it saw last year will depend on the future of generative AI and how well it can monetize it.
Currently, anyone can access ChatGPT for free, but a $20/month subscription fee gives you access to the latest and greatest version, though individual subscribers aren't a real source of revenue for now.
Rather, it’s in enterprises. And at this point, no one knows how enterprises will integrate generative AI into their business models in the coming years.
For Nvidia's growth to be sustainable, the company needs to offer new software that is “so AI-hungry” that major companies that sell software to businesses, such as Salesforce and Oracle, sign annual contracts that give them access to the largest amounts of computing power, Neumann said.
“Otherwise, the central theme of building giant megawatt data centers packed with GPUs all over the world is somewhat risky.”
So should you buy Nvidia stock? That depends on how bullish you are about AI and its penetration into the economy.
“For better or worse, we believe Nvidia's future will be tied to the AI market for quite some time,” Collelo wrote.
