SoftBank Plunges 12%, SK Hynix Slides 8% as Asian Tech Selloff Tracks U.S. Losses on AI Cost Fears

Two Names, One Thesis

The scale of the individual moves signals where investor anxiety is concentrated. SoftBank Group, whose portfolio is among the most visible expressions of the global AI investment thesis, bore the worst of the session with a 12% decline. SK Hynix, the South Korean memory-chip manufacturer whose products sit at the foundation of AI compute stacks, lost 8%. That both names moved sharply in the same session, for the same stated reason, is less coincidence than confirmation: the market is reassessing the economics of the AI buildout as a single, unified trade.

Rising AI Infrastructure Costs Reframe the Sector Narrative

The catalyst is a growing unease over what it actually costs to build artificial intelligence at scale. Infrastructure spending that markets once priced as a durable growth driver is now attracting scrutiny over its near-term return profile. For SoftBank, any repricing of the AI build cycle hits directly at the core of its investment case. For SK Hynix, which supplies the high-bandwidth memory that large AI models require, weaker conviction about the pace of that buildout introduces demand risk — even if the long-term trajectory is left uncontested.

Asia Follows the U.S. Lead

The Asian technology rout did not form independently. The selling tracked losses in U.S. technology stocks, a pattern that has become characteristic of how AI-related sentiment travels across time zones. When U.S. markets sell on infrastructure cost concerns, Asian markets open with the same thesis already repriced elsewhere and little structural reason to hold. The cross-Pacific correlation leaves Asia-listed technology stocks with a narrow buffer against negative U.S. sessions, particularly when the bearish catalyst is global rather than company-specific. For portfolio managers running cross-listed tech exposure, the message from this session is that there is currently nowhere in the trade to hide.