AI chip race: Fears grow of huge financial bubble


OpenAI Sam Altman, right, discusses the need for more chips designed for artificial intelligence with Intel CEO Pat Gelsinger on February 21, 2024, during a conference in San Jose, California. 

OpenAI Sam Altman, right, discusses the need for more chips designed for artificial intelligence with Intel CEO Pat Gelsinger on February 21, 2024, during a conference in San Jose, California. 
| Photo Credit: AP Photo/Michael Liedtke

Sam Altman caused more than a stir in early February when he called for a $5 to 7 trillion (€4.65 to 6.5 trillion) global investment to produce more powerful chips for the next generation of artificial intelligence platforms. Many industry analysts were left open-mouthed at the figure cited by the OpenAI chief executive, which is equivalent to almost a quarter of the US federal budget.

Altman wants to solve some of the major issues faced by the AI sector, which includes a major shortage of chips and semiconductors needed to power large language models like his firm’s ChatGPT, the Wall Street Journal reported in early February. The US entrepreneur has warned that vastly more powerful computing will be needed to help AI eventually overtake human intelligence. Altman recently held discussions with potential investors in the United Arab Emirates, the business daily said.

Unprecedented investment demands

“Asking for $7 trillion is just indecent,” Pedro Domingos, Professor Emeritus of computer science and engineering at the University of Washington, told DW. “It is an order of magnitude more than the entire chip industry has spent in its history.” Domingos said Altman would likely settle for around $700 billion in backing, which is still a far cry from the value of the entire AI chip sector. Canadian-Indian analytics firm Precedence Research recently calculated the industry could be worth some $135 billion by 2030.

Also Read | Sam Altman: Back at OpenAI, but questions on initial firing and future of AI persist

Others think that Altman’s projection might not be so far out if the ambition is for AI to eventually become smarter than humans in every way. “Right now, ChatGPT4 is only text,” Dylan Patel, chief analyst at SemiAnalysis, told DW. “But what if you add images, video, audio, and motorised tactile feedback? And what if we assume that AI does outpace humans on all fronts? That is going to cost hundreds of billions or even trillions of dollars.”

In the latest sign of the speed that AI is progressing, OpenAI last week unveiled a platform called Sora, for creating high-quality short videos from a simple line of text.

AI chip race heats up

Before Altman’s projection was made public, the world’s major governments—the United States, China, Japan, and several European countries—were already trying to secure or maintain a share of the chip industry for themselves.

Over the past 18 months, Washington has also levied sanctions on Beijing to stop Chinese firms from gaining access to US-designed chips. But rather than hobble Beijing’s ability to develop advanced AI computing power, Domingos said the sanctions were “counterproductive.” “There are many ways that China can obtain US chips through intermediaries. But those sanctions also encourage China to develop its own capacity and be less reliant on US chips,” the author of the book The Master Algorithm said. Indeed, the US sanctions have emboldened Chinese leaders who have pledged to step up their investments in AI chip production.

China catching up fast

“China is subsidising AI chips to the tune of $250 billion over the next decade to build a manufacturing supply chain and catch up,” Patel noted. He said China is currently about four to five years behind Taiwan, the global leader in chip manufacturing, and two to three years behind in semiconductor design—a race currently being won by US chip firms.

Also Read | Why did ChatGPT creator OpenAI oust its CEO Sam Altman?

Other countries may struggle to enter the AI chip-producing ring, as they do not have huge tech firms to commit tens of billions of investments, like Microsoft—which backs Altman’s OpenAI and Google, which unveiled its own AI chip in 2023. “If Germany wants to be a leader in AI, they are going to have to subsidise it because the likes of Mercedes Benz and Daimler are not necessarily going to invest a ton on advanced chips,” said Patel.

Advanced chips a ‘strategic commodity’

Economic historian Chris Miller, the author of the book Chip War, told DW that more countries have realised that ultra-high-speed chips have become a “strategic commodity,” amid the current geopolitical standoff between world powers. He predicted that the US government and others “will be quite sensitive about where the chip plants are located and who is involved in their production” to avoid autocratic countries like China from using AI for nefarious purposes.

NVIDIA leads stock market melt up

NVIDIA is the market leader in AI chip design. The Santa Clara, California-based firm is now valued at $1.8 trillion, making it the third-largest company on the US stock market, trailed by the likes of AMD and Intel. Amid a stock market melt-up—a period of rapid market growth fueled by investor optimismNVIDIA has seen its value rise by $296.5 billion in January, which most analysts think is unsustainable.

Domingos likened the current investor craze for AI to a “balloon that is inflating very rapidly,” until it bursts. “A lot of people, companies, countries are going to lose a ton of money. There is going to be a lot of carnage,” he told DW. “But in the longer term, AI will be like the Internet. Who cares about the dotcom bust these days? The Internet is a reality, it is all-pervasive and the basis for the next advancement in technology.”


Source link