The semiconductor cycle entered a critical inflection point this week. SK Hynix’ record $29 billion American Depositary Receipt offering—the largest foreign IPO in history—is set to list on the NASDAQ Friday under ticker SKHY, paired with Samsung’s simultaneous domestic capex blitz. Combined, the two Korean memory giants have committed 2,000 trillion won (approximately $1.3 trillion USD) to new semiconductor fabs, advanced packaging facilities, and extreme ultraviolet lithography equipment through 2028.

For SK Hynix specifically, the $29 billion raise will be allocated entirely to domestic production: 31 trillion won ($20.2B) for the Yongin Semiconductor Cluster’s flagship fab, 19 trillion won ($12.4B) for advanced packaging in Cheongju, and 12 trillion won ($7.8B) for EUV equipment acquisition. The strategic timing is deliberate. AI-driven memory demand has created unprecedented supply constraints: HBM (high-bandwidth memory) orders are sold out through 2027, with SK Hynix commanding 60% of that market. Memory pricing surged 44% for DRAM and 53% for NAND in the second quarter alone—a multi-decade high that has justified the investment thesis.

Yet the NASDAQ listing also signals something else: management de-risking at peak visibility. SK Hynix currently trades at 6.2 times forward earnings compared to Micron’s 7 times—a valuation discount despite superior HBM positioning and 70%+ operating margins. By listing in the US, SK Hynix removes that discount friction and monetizes at the precise moment when global institutional capital is most aggressive on the AI trade. The message is implicit: *the window for capital raising at peak prices is now.*

The Bitcoin/Tesla short-seller has pivoted to explicit bearishness on semiconductor valuations, entering what market observers describe as “viper mode.” Burry is currently shorting the SOXX (Philadelphia Semiconductor Index) ETF and NVIDIA via put options, with a singular thesis: Korea’s capacity expansion marks “the beginning of the end” for the chip cycle as we know it.

Burry’s argument inverts the consensus narrative. While markets celebrate “insatiable AI demand justifies infinite capex,” the historical cycle pattern shows that conviction peaks precisely *before* supply-driven deflation begins. Memory margins at 70% are extraordinary because supply is constrained. By 2027–2028, when Korean and Samsung fabs come online, those margins compress to 45–50%, a structural deterioration that renders today’s valuations unsustainable.

The positioning shift is meaningful. When a credible bear who has been correct on structural AI infrastructure themes enters significant short positions at precisely the moment consensus is most bullish, it signals the insider market is repricing risk. Burry’s puts are expensive—the market is not taking him seriously—but his willingness to size into the trade at elevated cost suggests conviction on timing.

Samsung Electronics reports Q2 2026 earnings this Tuesday (July 7). The market is modeling an 18-fold year-on-year increase in operating profit, with operating margins exceeding 50%—a level not seen since the 2017 memory boom. This earnings print will determine whether the Korea capex narrative holds (management affirms capacity deployment and demand resilience) or cracks (guidance moderates on pricing pressure or demand weakness).

Samsung’s positioning is strategic. Unlike SK Hynix, which is using IPO capital to fund Korean fabs, Samsung is funding capex from operations—signaling confidence in earnings durability. But Tuesday’s guidance will reveal whether management believes those margins persist or faces compression. Any commentary hinting at moderating capex or “cautious demand outlook” is the first crack in the cycle’s consensus.

Korea’s $1.3T capex expansion creates outsized opportunity for Taiwan-listed semiconductor equipment and substrate suppliers, Japanese fab-equipment makers, and Vietnam/Malaysia assembly partners.

Taiwan Exposure: TSMC (2330.TW) benefits indirectly—advanced-node demand (N3/N4) for AI chips remains strong through 2027, insulating foundry margin pressure. But if Burry’s deflation thesis plays out and AI infrastructure investment cools, foundry orders typically peak 12 months prior. Watch TSMC guidance for any softening in 2027 demand outlook.

Japan & Equipment Cycle: Tokyo Electron (8035.T) and Shin-Etsu Chemical (4063.T) are the pure-play capex accelerators. Korean fab construction peaks in 2H 2026 and early 2027; orders spike now, but execution is a 12-month lag. Both names benefit from the Samsung/SK Hynix spending wave.

Vietnam & Malaysia Assembly: Samsung Vietnam and Intel Penang face near-term tailwinds from Korean packaging fab construction. GlobalFoundries Singapore (Mubadala-majority owned) is a secondary play if TSMC advanced-node demand cools, but remains exposed to secondary chiplet/substrate demand.