SK Hynix to close $28 billion ADR bookbuild on Wednesday after oversubscription, source says

SK Hynix is set to close ADR bookbuilding for its $28B offering on Wednesday after orders already covered the deal multiple times, with major demand from U.S.-based investors (chunky orders starting around $200M and >$1B for larger bids). The ADR sale—17.79M new shares, priced off a 242,500 won reference per ADR with 10 ADRs per common share—highlights continued appetite for exposure to a key AI-memory supplier. Despite the strong demand, SK Hynix shares fell as much as 3.59% Wednesday as tech stocks wobbled amid renewed Middle East conflict.

Analysis

The important signal is not the capital raise itself; it is that the market is still willing to absorb a very large supply block in a volatile tape without demanding a clear distress discount. That usually marks an environment where AI infrastructure is still treated as a scarcity trade, which is constructive for NVDA and, more indirectly, GOOGL because it lowers the odds that memory availability becomes the bottleneck to accelerator deployment or cloud AI rollouts.

Second-order, the real loser is the thesis that AI hardware spend is constrained by financing rather than demand. If HBM capacity can be funded and expanded, it supports higher unit shipments and reduces the chance of near-term component shortages, which should help hyperscalers convert capex into revenue more cleanly. The counterpoint is that this can also soften the scarcity premium in the memory stack over 6-18 months if supply growth outruns the training wave.

Near term, the key falsifier is aftermarket weakness or a wide pricing concession, which would suggest the book was momentum-led rather than fundamental. Over the next 1-3 months, watch HBM lead-time commentary and cloud capex updates: any downgrade there would reverse the read-through quickly. The consensus may be overrating the breadth of the signal; this is supportive for the AI complex, but it is not a blanket buy signal for every tech beta name, especially lower-quality China exposure like BABA.

Key Decisions for Investors

• Maintain/add NVDA on 2-3% pullbacks over the next 1-3 weeks; if using options, prefer 6-10 week call spreads rather than outright calls to limit premium decay. Falsifier: weak HBM pricing commentary or a material guide-down from major AI customers.
• Keep GOOGL overweight as a second-order beneficiary of improved AI supply-chain visibility; use upcoming earnings and cloud capex commentary over the next 1-2 months as the confirmation point. Risk/reward is favorable as long as AI spend is translating into product and cloud revenue.
• Do not chase BABA on this theme. Treat any sympathy move in China tech as low-conviction beta rather than direct exposure to the AI supply-chain trade; fade strength if it is not accompanied by a broader China policy re-rating.
• Set an alert on SK Hynix aftermarket pricing and the first post-listing trading sessions: if the ADR prices poorly or underperforms the sector for 2-3 days, reduce AI-beta exposure because it would signal the demand signal was more order-flow than fundamental.
• Watch for a 1-3 month reversal signal in memory and hyperscaler capex. If HBM lead times shorten materially or cloud capex guidance flattens, trim NVDA/GOOGL exposure into strength because the scarcity premium in the supply chain will start to compress.