How Nvidia became the powerhouse chipmaker of the AI craze

Nvidia has surged to new highs on the stock market in recent weeks, fueled by optimism about the future of artificial intelligence (AI).

The chipmaker — which is a leading manufacturer of the graphic processing units (GPUs) typically used in AI — closed above $2 trillion for the first time Friday, making it the third most valuable company on Wall Street, behind only Microsoft and Apple.

With the expansive “moat” Nvidia has secured through years of investment and the development of its own widely used software ecosystem, experts said it is unlikely competitors will be able to span the gap any time soon.

“I think it’s Nvidia’s game to lose, and they’re not showing any signs of losing it right now,” Stacy Rasgon, a senior analyst at Bernstein Research, told The Hill.

Nvidia has been developing GPUs for decades. The chips were primarily used for video games, until a discovery a decade ago prompted the machine learning community to begin using GPUs, said Tianqi Chen, an assistant professor in Carnegie Mellon’s Machine Learning Department.

Computer scientist Geoffrey Hinton, who is known as one of the “godfathers” of AI, found that GPUs were more efficient at the kind of large-scale computing required for machine learning, leading to the start of a “deep learning revolution,” Chen said.

“The machine learning community started to embrace GPU computing,” Chen said. “As of today, right, almost all the AI models that run deep neural learning networks … a lot of them, even a majority, runs on GPU.”

Nvidia noticed this development and began creating libraries for machine learning within its software ecosystem, called CUDA, Rasgon said.

While Nvidia was focused on developing its AI capabilities, its primary competitor in the GPU market, Advanced Micro Devices (AMD), fell on hard times.

“It’s only in recent years that AMD and even others have had the resources in place to start investing in data center and AI with GPUs,” Rasgon said. “But by then, Nvidia’s had like a 10-year lead.”

“A lot of it comes down to this: They had an early recognition that this was going to be important. They started dedicating resources to develop products, both hardware and software, to go toward this. They had competitors that were either not interested in doing this or not capable of doing it. And they never lost faith,” he added.

Nvidia’s expansive lead on GPUs is bolstered by the existence of its own software ecosystem.

“Nvidia took an early lead in AI GPU hardware, but more important, developed a proprietary software platform, Cuda, and these tools allow AI developers to build their models with Nvidia,” Morningstar equity strategist Brian Colello said in a recent report.

“We believe Nvidia not only has a hardware lead, but benefits from high customer switching costs around Cuda, making it unlikely for another GPU vendor to emerge as a leader in AI training.”

If a developer were to try to switch to AMD or Intel parts, they would have to entirely rewrite their code, Rasgon noted.

“It’s an enormous undertaking,” he said. “And time is money, right? I mean, you want to get to market with this stuff as quickly as possible. It’s much easier just if you’ve developed everything over the last 10 years on Nvidia parts to just keep using them.”

Nvidia has begun to see its investment in GPUs pay off over the past year, after the launch of OpenAI’s popular ChatGPT tool sparked fierce competition among major tech companies to develop and release their own generative AI models.

The chipmaker achieved a $1 trillion market value for the first time in May 2023. Nvidia has continued its upward climb in recent months, with its shares up 77 percent since the start of the year.

Nvidia initially crossed the $2 trillion mark late last month, after the company posted strong fourth quarter results that beat expectations. It added $277 billion in market value in one day to briefly take it above $2 trillion, breaking Wall Street’s record for the largest one-day gain.

Nvidia’s shares surged once again last week to close above $2 trillion for the first time, after Dell posted stronger than expected fourth quarter results. Dell uses Nvidia’s GPUs in its servers, according to Reuters.

“Accelerated computing and generative AI have hit the tipping point. Demand is surging worldwide across companies, industries and nations,” Jensen Huang, the founder and CEO of Nvidia, said in the company’s latest earnings report.

The company took a hit in China last year, with its data center revenue declining “significantly” in the region after the Biden administration placed restrictions on the export of advanced chips in late 2022.

Nvidia developed two new chips with reduced capabilities to bypass the restrictions, but the administration ultimately cracked down on these chips as well in October 2023, citing concerns that American technology could be used to strengthen the Chinese military.

The dominance of the Santa Clara, Calif.-based company appears to be largely unshakeable, at least at the current moment.

Fellow chipmakers such as AMD and Intel seem unlikely to pull ahead of Nvidia, and in-house options from companies including Google, Microsoft, Amazon and Meta could serve specific purposes but would likely lack the flexibility offered by GPUs, Rasgon said.

Chen suggested it is possible for alternative chips to gain some of the market share if companies invest heavily in the software component. However, he also said he doesn’t think Nvidia will lose its leadership position.

“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,” Morningstar’s Colello added.