A trader from Goldman Sachs in Japan has issued a warning regarding the recent sharp correction in tech stocks, indicating that the memory chip and AI infrastructure sectors are facing structural pressures rather than mere short-term volatility.
According to Goldman Sachs trader Ippei Yamaura in a recent report, tech stocks experienced a record 4-standard-deviation sell-off last week, with the trigger traceable to initial declines in South Korean memory stocks. While Micron's earnings report briefly boosted market sentiment, this rebound failed to hold into Friday—as reports emerged that OpenAI is considering delaying its IPO until 2027 while also lowering its $1 trillion valuation target, unsettling market confidence once again.
The market impact is already evident: AI infrastructure stocks are broadly under pressure, while hyperscale cloud providers and Apple are showing relative resilience. Yamaura believes the market is starting to price in a "peak in memory supply tightness." Concurrently, reports indicate that Apple is lobbying the U.S. government for approval to source memory chips from Chinese semiconductor firm ChangXin Memory Technologies (CXMT) to alleviate cost pressures from rising memory chip prices. If approved, this would represent a material challenge to Micron.
OpenAI's Price Cuts and IPO Uncertainty: The Spark for Market Concerns
Yamaura notes in the report that a deeper trigger for this tech stock correction was likely OpenAI's price cut announcement on June 11—a move intended to compete with Anthropic. For investors, this action implies a further delay in the timeline for OpenAI's business to become self-sustaining and move away from continuous cash burn.
Simultaneously, OpenAI's IPO prospects have grown more complex. Recent media reports suggest OpenAI is considering delaying its listing plans, citing market instability following the cold reception to SpaceX's IPO and uncertainty around achieving a $1 trillion valuation target. As of the end of March, OpenAI's latest valuation was $825 billion.
The deeper market logic is that the combination of price cuts and a delayed listing has resurfaced a familiar, pressing question: Can AI investments deliver sufficient returns? This doubt, in turn, suggests the threat to software companies may not be as severe as previously feared, and the current pace of AI infrastructure investment may be unsustainable.
Three Core Risks: Goldman's Complete Warning for Micron
Yamaura's report clearly outlines three major downside risks flagged by Goldman for Micron and the broader memory sector.
First, a slowdown in HBM price momentum. As industry capacity rapidly expands through fiscal years 2027-2028, the price support for High Bandwidth Memory (HBM) will be tested.
Second, erosion of DRAM market share. The rise of Chinese manufacturers is creating pricing pressure. Reports indicate Apple is seeking U.S. government approval to purchase DRAM from ChangXin—Apple had previously raised iPhone prices due to rising memory costs and is now evidently unwilling to continue bearing this expense.
Third, a broad, sudden deceleration in AI server investment. Goldman points out that even if the U.S. government rejects Apple's procurement request, similar dynamics are unfolding broadly—Qualcomm is reducing its reliance on HBM, and Nvidia is working to lower its memory usage.
Furthermore, downstream customers are actively seeking ways to reduce their dependence on memory chips, a trend Yamaura notes has strong momentum for acceleration, despite Micron's view that supply shortages will persist.
From Structure to Landscape: AI Competition Intensifies, Memory Demand Narrative Weakens
From a broader perspective, Yamaura highlights a noteworthy market phenomenon: Chinese startups and small-to-medium enterprises are using a combination of multiple AI models to avoid high AI usage fees—a clear departure from the commercial expectations of leading platforms like OpenAI. Competition among AI models has intensified even before OpenAI's potential listing.
These dynamics collectively point to one conclusion: market confidence in the core narrative of "persistent memory supply tightness" is wavering, and the recent divergence between AI infrastructure stocks and hyperscale cloud providers is a direct manifestation of this repricing.
On strategy, Yamaura states that given last week's severe tech stock volatility, Goldman is adjusting its near-term tactical stance toward a more defensive posture. At the same time, renewed geopolitical risks following U.S. military action against Iran make cyclical sectors difficult to approach in the short term as well.
However, Goldman also points to a key support: the U.S. economic fundamentals overall remain robust. Therefore, Yamaura believes any shift toward defensive plays or index-level selling should be viewed as cyclical rather than trend-driven. For stock-specific actions, Yamaura offers clear directional advice for medium-to-long-term investors: for stocks where momentum has broken, the inclination should be to buy on dips rather than aggressively reduce positions.
Comments