Korean chip stocks will win no matter who survives the AI giants' 'deathmatch': KB Financial Group

The important second-order read-through is that the current rally in Korean memory is still being treated as a cyclical trade by foreign fast money, while local strategists are framing it as an earnings reset with multiple years of runway. That mismatch usually persists until sell-side consensus has fully revised peak-to-trough margins and capex assumptions, which means the biggest upside may still be in the next 2-3 earnings revisions rather than the next headline. If this is right, suppliers with leverage to DRAM/HBM tightness should continue to outperform the broad KOSPI even if index-level flows stay negative.

The labor-union angle is the near-term wildcard because it can hit the market through a different channel than investors are focused on: not just lost output, but a risk premium on execution. Even a short-lived work stoppage can matter if inventories are lean and lead times are stretched, because it converts pricing power into delivery uncertainty and can slow customer qualification for next-gen products. The more interesting implication is that any disruption would likely be used by global buyers to diversify sourcing, which could benefit secondary foundry and memory alternatives over a 6-12 month horizon.

The foreign selling matters less as a signal on fundamentals and more as a setup for squeeze dynamics. When a market is being sold by non-domestic institutions while domestic narratives remain constructive, upside often accelerates on a single catalyst: stronger guidance, a positive AI server order comment, or evidence that margins are extending rather than peaking. The consensus is likely still underestimating how long the memory cycle can stay tight if AI-related HBM demand remains the marginal buyer and foundry utilization normalizes slower than expected.