At Marketplace, we say all the time that the stock market is not the economy. But the stock market does sometimes tell you where the economy might be headed. 

Case in point, Nvidia, the computer chip designer, has seen its stock price more than double since the start of the year, to about $300 a share. The Silicon Valley-based company now has a bigger market capitalization than Visa, Tesla or Facebook.

The reason investors have been so bullish on Nvidia? Two letters: AI.

Ask ChatGPT for a short essay on whether a hot dog is a sandwich, and it’ll write eight wholly original and pretty entertaining paragraphs in seconds. And behind each of those paragraphs is a system of math.

“It was something like 400 quadrillion operations. It’s a lot,” said Stacy Rasgon, an analyst at Bernstein Research.

ChatGPT needs really powerful computer chips to run all those operations behind its responses, not to mention to train the AI in the first place. Right now, not a lot of companies make those chips.

“Clearly the one that’s leading is Nvidia, and then everybody else is far, far behind,” Rasgon said.

Companies like Intel specialize in traditional CPUs, the chips that power PCs and take on one computing task at a time, sequentially. Nvidia specialize in GPUs, or graphics processing units. They’re typically used for video games, but they’re also multitasking wizards.

“When you’re dealing with AI workloads, you require a ton of parallel calculations,” explained Vivek Arya, an analyst with Bank of America. “And that is why, you know, customers don’t mind paying extra for those GPUs.”

Nvidia’s latest AI GPUs cost about $30,000 a chip. Arya estimates the total market will reach about $80 billion in four years. Other semiconductor companies are trying to get in on the AI business, and Google and Amazon have also launched their own chip lines.

But Karson Elmgren at Georgetown University’s Center for Security and Emerging Technology said it’s important to remember that Nvidia and the rest of those companies only design the advanced chips that power AI.

“They outsource all their manufacturing to TSMC, the Taiwan Semiconductor Manufacturing Co., which is based in Taiwan,” he said.

That makes the entire AI industry vulnerable to supply chain hiccups and geopolitical threats.