SK Hynix Shares Crash Despite Record Q2 Earnings as AI Memory Shipments Miss Elevated Expectations

The Quick Take

• SK Hynix shares crashed despite posting record Q2 2026 profit and revenue, as earnings missed analyst expectations set on aggressive AI memory demand assumptions
• AI memory shipment volumes came in below projections, raising questions among investors about the sustainability of the current AI hardware investment cycle
• The selloff triggered a broad semiconductor sector correction and weighed heavily on South Korean markets, with KOSPI trading halts occurring amid intensified foreign selling

SK Hynix delivered record profit and revenue for Q2 2026, yet saw its shares plunge in one of the more dramatic post-earnings reactions in recent memory for the semiconductor sector. The paradox is explained by the gap between what analyst models had priced in and what the company actually reported: expectations had been calibrated to an accelerating AI memory demand scenario where high-bandwidth memory chip shipments would show pronounced sequential growth. Instead, the actual shipment volumes and forward-looking commentary from management painted a more measured picture of AI hardware demand ramp timing.

The core concern among investors is whether the current AI infrastructure investment cycle is beginning to moderate more quickly than the semiconductor supply chain had anticipated. SK Hynix had been positioned as one of the primary beneficiaries of AI data center buildout through its high-bandwidth memory products, which are essential components in the AI accelerator chips used for large-scale model training and inference. A shortfall in HBM memory demand therefore reads as a signal about the broader pace of AI capacity expansion, with knock-on implications for the entire semiconductor hardware supply chain.

The market reaction extended well beyond SK Hynix, triggering a broader Asian technology sector selloff and contributing to KOSPI trading halts as foreign fund managers reduced their South Korean technology exposure. The broader context is that semiconductor stocks had been pricing in a sustained AI demand acceleration, and any evidence of demand moderation tends to trigger outsized stock reactions given the elevated multiples at which the sector was trading. Analysts across the semiconductor coverage universe are now revisiting their AI memory demand models and shipment forecasts for the second half of 2026, which will determine whether the SK Hynix earnings miss was an isolated result or an early signal of a broader industry adjustment.

Synthesized from 1 source.