AI Stocks Plunge Violently, Is It GLM's DeepSeek Moment?

marsbitPublished on 2026-06-24Last updated on 2026-06-24

Abstract

On Tuesday, AI-related stocks faced their most severe sell-off of the year, triggered by a sharp decline in South Korean equities. The plunge, led by key AI supply chain players Samsung and SK Hynix, spread to US markets, heavily impacting semiconductor, memory, and storage stocks. Analysts linked the downturn to the recent release of China's powerful open-source model, GLM-5.2, drawing parallels to the market shock caused by DeepSeek's release earlier in 2025. This event reignited investor concerns that cheaper, competitive AI models could undermine the economic rationale for the massive data center investments by major US tech firms. The sell-off was seen as a repricing of the AI trade rather than a rejection of AI demand itself. Market focus shifted from whether AI will grow to whether the current valuations justify the enormous capital expenditures. While some strategists view this as a necessary correction after excessive gains, questions remain about the sustainability of financing for AI infrastructure, particularly as companies increasingly rely on debt. All eyes are now on upcoming earnings reports, like Micron's, to gauge the hardware sector's health. The core debate has become about the cost of growth and which companies can successfully convert their massive investments into sustainable cash flow.

On Tuesday, artificial intelligence trading faced its most intense stress test of the year. It started with a sharp decline in the South Korean stock market led by Samsung Electronics and SK Hynix, before the sell-off spread to the US trading session, hitting memory, storage, and semiconductor stocks particularly hard.

The South Korean KOSPI index tumbled nearly 10% at one point on Tuesday, triggering a 20-minute trading halt. Samsung Electronics and SK Hynix, core players in the global AI supply chain, both suffered heavy losses. According to The Wall Street Journal, the downturn subsequently spread to US stocks, with the Nasdaq Composite closing down 2.2% and the S&P 500 falling 1.4%. AI and chip-related stocks led the decline as investors began reassessing the uncertainties surrounding data center construction costs and future revenue realization.

Some view this global AI stock sell-off as "GLM's DeepSeek moment," mirroring the AI stock plunge triggered by DeepSeek's release in early 2025, where an overly powerful open-source model sparked skepticism about US AI dominance. Investment bank Jefferies noted in a report that GLM-5.2 has entered the top three in the global large language model rankings.

Nathan Lambert, a senior research scientist at Allen Institute for AI and author of Interconnects, called it a "step change" for open-source agent models and compared the market reaction to the shock caused by DeepSeek R1 in early 2025. This discussion within tech circles was quickly picked up by financial media. Barron's interpreted Tuesday's tech stock decline as the return of "cheap Chinese AI" concerns, agreeing this drop is a repeat of January's DeepSeek shock. Gavekal Research analyst Will Denyer was quoted saying GLM-5.2 is one of the most impressive Chinese challenges to US AI dominance yet. For investors, the issue is not just stronger Chinese models, but whether the current valuations of US tech giants, supported by hundreds of billions in data center spending, can hold if cheaper, open-source models become good enough.

Arun Sai, senior multi-asset strategist at Pictet Asset Management, told the Financial Times that the market is facing two pressures simultaneously: growing doubts about AI investment returns and rising interest rate expectations due to US economic resilience. Ben Inker, co-head of asset allocation at GMO, also noted that these stocks had risen too much and were due for a pullback.

The extent of the decline in US chip stocks suggests money is not fleeing the entire tech sector, but rather the hardware chain that previously benefited most from the AI infrastructure narrative. Eric Johnston, chief equity and macro strategist at Cantor Fitzgerald, summarized current trading as selling "the companies spending the most money," pointing to hyperscalers like Alphabet, Amazon, and Meta, which still plan to invest hundreds of billions in AI data centers.

The South Korean market decline might be more attributable to specific events. Lee Chan-jin, head of South Korea's Financial Supervisory Service, said on Monday that the country's previous approval of leveraged single-stock ETFs related to Samsung and SK Hynix was too hasty. The market was also hit by MSCI's decision not to include South Korea on its watchlist for developed market status, dashing investors' expectations of passive fund inflows for now.

Sell-side analysts are focusing on Micron's upcoming earnings report. Dilin Wu, strategist at Pepperstone Group, told Bloomberg that Micron's report this week will be a key test for hardware chain sentiment; strong results would directly benefit Samsung and SK Hynix. Lee Jae Mahn, a strategist at Hana Securities in Seoul, noted that SK Hynix's rally relative to Samsung has been too fast, reflecting excessive optimism.

Another unsettling variable for the market is the increasing reliance on debt to finance AI infrastructure. The Guardian cited Ipek Ozkardeskaya, senior analyst at Swissquote, who noted that SpaceX's pursuit of large-scale debt financing soon after its IPO has renewed investor concerns about major tech companies over-investing in AI infrastructure and supporting this race through debt.

However, bulls are not ready to declare the AI trade over. Dan Ives, global head of technology research at Wedbush Securities, stated in a Tuesday report that the South Korean market pullback would pressure US tech stocks, but he still believes the AI revolution is in its early stages, calling this more of a "gut check" for the tech trade. Jonathan Schiessl, deputy chief investment officer at Westminster Asset Management, also termed this decline a necessary correction after overheating, not the end of the story.

This suggests Tuesday's drop is more a market repricing of the AI trade rather than a denial of AI demand itself. The core question shifts from "Will AI grow?" to "Is the price paid for that growth too high?": Who can convert capital expenditure into cash flow, whose valuation is already overstretched, and who will be forced to sell when leverage and crowded trades unwind.

Trending Cryptos

Related Questions

QWhat triggered the sharp sell-off in global AI-related stocks, particularly in South Korea and the US, as described in the article?

AThe sell-off was triggered by a combination of factors. In South Korea, it was linked to a regulatory comment from the Financial Supervisory Service about the hasty approval of leveraged single-stock ETFs for Samsung and SK Hynix, and disappointment over MSCI not adding South Korea to its watchlist for developed markets. In the US, the decline was driven by market concerns sparked by the release of China's powerful open-source AI model, GLM-5.2, which led to a reassessment of the costs and future returns of massive AI infrastructure investments by US tech giants, reminiscent of the impact of DeepSeek R1 in early 2025.

QHow do analysts cited in the article compare the impact of GLM-5.2 to the DeepSeek R1 event?

AAnalysts and researchers directly compare the two events, labeling the market reaction to GLM-5.2 as a 'DeepSeek moment' or a 'replay' of the DeepSeek impact. They describe GLM-5.2 as a 'step change' for open-source agent models and view it as China's most impressive challenge yet to US AI dominance. The core parallel is the concern that sufficiently powerful and cheaper open-source models could undermine the economic rationale for the massive capital expenditures planned by US hyperscalers.

QAccording to the article, what is the main concern for investors regarding large US tech companies' massive AI infrastructure spending?

AThe main concern is whether the thousands of billions of dollars in planned data center spending by US hyperscalers like Alphabet, Amazon, and Meta can be justified and support their current valuations if cheaper, open-source AI models (like GLM-5.2) prove to be 'good enough.' Investors are questioning if these enormous capital expenditures will translate into expected future cash flows and returns.

QWhat specific event or report are market participants watching as a key test for the hardware supply chain's health after the sell-off?

AMarket participants are closely watching the upcoming earnings report from Micron Technology. Analysts cited in the article state that Micron's financial results will serve as a crucial indicator for the health of the hardware supply chain. A strong performance is seen as potentially benefiting related stocks like Samsung and SK Hynix.

QDespite the sell-off, what is the prevailing view among bullish analysts about the long-term AI trend?

ABullish analysts do not believe the AI trade is over. They characterize the sell-off as a necessary correction after an overheated rally, a 'gut check,' or a market repricing rather than a story-ending event. They maintain that the AI revolution is still in its early stages. The core question has shifted from whether AI will grow to whether the price paid for that growth is too high.

Related Reads

Goldman Sachs Research Report Analysis: Circle and USDC Are Moving Beyond the Crypto World, Cross-Border Payments and AI Agents Become New Battlegrounds

Goldman Sachs published a summary of its meeting with Circle Internet Group (issuer of USDC) on July 5th. The core takeaway is that stablecoins, led by USDC, are evolving from a crypto-native tool into foundational infrastructure for traditional finance and the AI economy. USDC's use cases are rapidly expanding beyond crypto trading into cross-border payments, e-commerce, capital market settlements, and notably, payments for AI agents. Circle's management emphasized that stablecoin growth is now decoupled from crypto market cycles, driven by this diversification. They outlined five key application layers and highlighted USDC's network effects, global liquidity depth, and regulatory compliance as competitive moats. Circle distinguishes USDC from bank-issued tokenized deposits, arguing the former is an open, internet-native system without bank credit risk. Strategically, Circle is building a broader fintech platform with its Arc Layer 1 blockchain, the Circle Payments Network for cross-border transfers, and an "Agentic Stack" to serve AI agent economies, where USDC already dominates. Regarding regulation, Circle views potential U.S. legislation like the CLARITY Act as a catalyst for growth rather than a constraint, expecting it to encourage broader institutional adoption and active usage. Goldman Sachs maintains a Neutral rating on Circle with a $96 price target, noting the company's shift from a pure stablecoin issuer to a financial infrastructure provider. Key risks include competition from USDT and potential earnings pressure from declining interest rates on its reserve assets.

marsbit35m ago

Goldman Sachs Research Report Analysis: Circle and USDC Are Moving Beyond the Crypto World, Cross-Border Payments and AI Agents Become New Battlegrounds

marsbit35m ago

The Robinhood Stock Tokens You Bought Are Just Debts from Jersey Island

The Robinhood stock tokens you buy are essentially debt securities issued by a shell company in Jersey, not real equity. These tokens merely track stock prices like NVIDIA or Apple but grant no shareholder rights like voting or dividends. If the underlying company fails, you have no claim on its assets. Instead, you hold a debt instrument from Robinhood Assets (Jersey) Limited, which promises returns based on stock performance. If this Jersey entity goes bankrupt, you become an unsecured creditor. This complex structure stems from Robinhood's past crisis during the 2021 GameStop short squeeze, where T+2 settlement caused liquidity issues. The blockchain-based tokens enable instant settlement, theoretically preventing such trading halts. The product is classified by the SEC as a "linked security" or structured note, carrying counterparty risk not borne by actual shareholders. It is available globally but excluded from the US, UK, and other major markets, while Robinhood offers a fully compliant, asset-backed token model in Europe under MiFID II. The system relies on oracles for pricing, which poses risks like manipulation and faulty liquidations seen in DeFi exploits. Robinhood profits from spreads and aims to become a full-chain settlement layer. Meanwhile, competitors like Ondo have launched SEC-registered, fully compliant equity tokens in the US with actual voting rights and dividends. Robinhood’s Jersey debt model appears as a transitional, regulatory-arbitrage product, aiming to capture market share ahead of future regulatory clarity.

Foresight News1h ago

The Robinhood Stock Tokens You Bought Are Just Debts from Jersey Island

Foresight News1h ago

Trading

Spot

Hot Articles

Discussions

Welcome to the HTX Community. Here, you can stay informed about the latest platform developments and gain access to professional market insights. Users' opinions on the price of AI (AI) are presented below.

活动图片