The introduction of ChatGPT has set fire to the stock prices of companies that make microchips, the brains of artificial intelligence. Betting on the potential of so-called generative AI is pouring in. The most eye-catching example of this rise is Silicon Valley’s NVIDIA, a top seller of chips used in artificial intelligence, whose stock has risen nearly 200 percent this year.
South Korean giant Samsung Electronics also wants to join the effort. Samsung, widely known for its consumer products, also has the world’s largest memory chip business and the world’s second-largest semiconductor foundry, which makes custom microchips for other companies.
Foreign investors have bought $8 billion worth of Samsung shares on the South Korean stock market this year, according to data provided by Hong Kong investment firm CLSA, which is already the largest annual purchase of Samsung shares by foreigners since 2000. is the maximum. The surge reversed a three-year plunge that saw foreign investors sell more shares in the company than they bought.
At an event in California last week, Samsung detailed what it calls its “vision for the AI age.” Samsung believes it can grab market share from major chip maker Taiwan Semiconductor Manufacturing Company, but recently the trend has reversed. TSMC enjoys about 60% of the total revenue of the global foundry business, compared with just 13% for Samsung, according to market research firm Counterpoint Research. The gap widened after 2021 as some of Samsung’s customers, including Nvidia, transferred their operations to the foundry business. TSMC.
Samsung said it spent $7.4 billion on its chip business in the first quarter of this year, when profits fell by a staggering 95%, some of which is expected to benefit the AI industry. The company is expanding production at a chip manufacturing complex in Pyeongtaek, about 60 miles south of Seoul, and at a chip factory in Texas. Samsung said it plans to work with the government on a $230 billion plan to build a “megacluster” of chip manufacturing in South Korea over the next 20 years.
Sanjeev Rana, senior analyst at CLSA, said the optimism was tied to Samsung’s memory chip business, which typically accounts for about half of the company’s operating profit.
Compared to traditional servers (the hardware that powers desktops and databases), servers built for artificial intelligence can require four times as much memory, called DRAM. Samsung holds about 45% of his DRAM market in the world. It is also the only major memory company to invest in more production, Rana added, despite falling memory prices across the industry.
The chip industry is known for its boom-bust cycles. After demand for memory chips surged during the pandemic, chip makers began their worst recession in years last fall. Samsung’s memory chip rivals, including U.S. Micron Technology Inc. and South Korea’s SK Hynix Inc., have said this year they will cut investment in production.
Some analysts believe Samsung’s spending during the downturn, thanks in part to artificial intelligence, will pay off in the long run as the memory division recovers.
“They will be very ready when demand returns,” said Rana.
But skeptics question whether Samsung can play the essential role in generative AI that it has played in smartphones and high-definition TVs. Last year, the company lost out when Nvidia chose SK Hynix as a supplier of high-power memory chips expected to become a burgeoning business area due to its prominence in future AI servers.
SK Hynix controls about 50% of the high-bandwidth memory (HBM) market, compared with Samsung’s 40%, according to market research firm Trendforce. SK Hynix shares are up more than 50% this year, surpassing Samsung’s 30% gain.
Samsung said it has already started supplying competing versions of HBM to “major customers.” The company added that its next-generation HBM will be launched this year.
Nam Hyun Kim, an analyst at equity research firm Arete Research, said the delay in Samsung’s HBM technology could be a symptom of a wider problem. In a February report, Kim wrote that Micron had also leapfrogged Samsung’s technology in DRAM and another type of memory, NAND flash.
“The problem with Samsung is that they always want to be big,” Kim said. “They’re spending a lot of money, but they’re no longer the technology leaders.”
Kim said Samsung should invest more in research and not worry too much about market share. “Samsung is a bigger player than Apple in smartphones,” he said. “But how many people think Samsung makes better smartphones than Apple?”
Samsung said in a statement that it has been successful in several aspects of its advanced semiconductor technology, allowing it to provide customers with “comprehensive solutions” in the evolving landscape of AI and other technologies.
Samsung’s own management offered a more sober diagnosis.
In May, Samsung’s semiconductor division president, Kyung Gye-hyung, admitted in a speech to college students that the company had been “lagging behind” TSMC by up to two years. This widely disseminated statement korean mediaThe approval was unusual for a company that has long prided itself on its technological leadership.
Kyung went on to assert that Samsung’s memory chips will be the “core” of AI supercomputers by 2028, saying, “We will outperform TSMC within five years.”
Jin Yoo Young Contributed to the report.
