On Tuesday, artificial intelligence trading faced its most severe stress test of the year. First, the South Korean stock market plummeted driven by Samsung Electronics and SK Hynix, followed by a sell-off during the U.S. trading session, with memory, storage, and semiconductor stocks taking a heavy hit.
The South Korean KOSPI index dropped nearly 10% at one point on Tuesday, triggering a 20-minute circuit breaker. Samsung Electronics and SK Hynix, as the core of the global AI supply chain, both saw sharp declines. According to The Wall Street Journal, this decline then spread to the U.S. stock market, with the Nasdaq Composite Index falling 2.2% and the S&P 500 Index dropping 1.4%; AI and chip-related stocks led the decline as investors began reassessing data center construction costs and the uncertainty of future revenue realization.
Some believe that the global AI stock downturn is a "Spectrum's DeepSeek Moment," a flashback to the early days of DeepSeek 25 years ago that caused a major AI stock crash, where this time an open-source model was deemed too powerful, causing market doubts about U.S. AI. Investment bank Jefurui stated in a report that Spectrum GLM-5.2 has now entered the top three global large model rankings.
Nathan Lambert, Senior Research Scientist at the Allen Institute for AI and author of Interconnects, referred to it as a "step change" in open-source intelligent agent models and likened the market reaction to the impact caused by DeepSeek R1 at the beginning of 2025. This technical discussion was quickly picked up by the market media. Barron's explained Tuesday's tech stock decline as a return of "Cheap China AI" concerns, likening this drop to a repeat of the January DeepSeek shock. Gavekal Research analyst Will Denyer was quoted as saying that GLM-5.2 is one of the most impressive challenges from China to U.S. AI dominance to date. For investors, the issue is not just that Chinese models are getting stronger, but whether cheaper open-source models are good enough, and if the hundreds of billions of dollars spent by U.S. tech giants on data centers can still support current valuations.
Arun Sai, Senior Multi-Asset Strategist at Pictet Asset Management, in an interview with the Financial Times, stated that the market is currently facing two pressures simultaneously: increasing doubt about AI investment returns and rising rate expectations due to the resilience of the U.S. economy. Ben Inker, Co-Head of Asset Allocation at GMO, also believes that the related stocks had previously experienced excessive gains and were due for a pullback.
The decline in U.S. chip stocks shows that the outflow of funds is not from the entire technology sector, but from the hardware chain that has benefited most from the AI infrastructure narrative. Eric Johnston, Chief Stock and Macro Strategist at Cantor Fitzgerald, summarized the current trading as selling off the "most expensive companies," pointing to Alphabet, Amazon, Meta, and others that still plan to invest billions of dollars in building AI data centers as hyperscalers.
The decline in the South Korean stock market may be more attributable to specific events. Financial Supervisory Service of Korea Director Lee Chan-jin stated on Monday that Korea's earlier approval of leverage single-stock ETFs related to Samsung and SK Hynix was too hasty. The market was also hit by MSCI's decision not to include Korea on its developed market watchlist, temporarily disappointing investors' expectations of passive fund inflows.
Sellers are now focusing on Micron's earnings report. Pepperstone Group strategist Dilin Wu, in an interview with Bloomberg, said Micron's earnings this week would be a key gauge of the hardware chain's strength; strong performance would directly benefit Samsung and SK Hynix. Lee Jae Mahn, a strategist at Hana Securities in Seoul, pointed out that SK Hynix's rapid rise relative to Samsung has been overly optimistic.
Another factor causing market unease is that AI infrastructure financing is becoming increasingly debt-reliant. The Guardian cited Swissquote's Senior Analyst Ipek Ozkardeskaya's view that shortly after going public, SpaceX sought significant debt financing, raising investors' concerns again about large tech companies overspending on AI infrastructure and starting to support this race through debt.
However, the bulls have not concluded that AI trading is over. Dan Ives, Managing Director of Global Technology Research at Wedbush Securities, stated in a report on Tuesday that the South Korean market's pullback would put pressure on US tech stocks, but he still believes that the AI revolution is in its early stages, more akin to a "gut check" in tech trading. Jonathan Schiessl, Deputy CIO of Westminster Asset Management, also referred to this drop as a necessary correction after overheating, rather than the end of the story.
This implies that Tuesday's decline is more like a market repricing of AI trading rather than a denial of AI demand itself. The real question has shifted from "Will AI grow" to "Is the price paid for growth too high": who can turn capital expenditures into cash flow, whose valuation is already stretched, and who will be forced to sell when leverage and crowded trades recede.
Welcome to join the official BlockBeats community:
Telegram Subscription Group: https://t.me/theblockbeats
Telegram Discussion Group: https://t.me/BlockBeats_App
Official Twitter Account: https://twitter.com/BlockBeatsAsia
