The Scarcity Trade

China Memory Capex go brrr Another 15B won order on June 1 for $025560.KS Mirae Corp Test handlers Q2 Started March 31st Since then they have announced 37B KRW in contracts so far in Q2 06/01 15.4B KRW (Yiling) 5/27/26 3.6B KRW (direct order from YMTC) 5/20/26 2.6B KRW (SK hynix) 5/20/26 4B KRW (SK hynix) 5/13/26 8B KRW (Unimos) 4/29/26 3.8B KRW (Yiling) *Unimos is assembly and test arm for YMTC *Yiling inferred to be a trading intermediary for YMTC and CXMT To put this into context - Full Year 2025 revenue was 50B KRW they have almost met that in one quarter Q126 - revenues were 21B KRW - EBIT was 4.9B KRW ~24% ebit margins They are about to hit 100% QoQ revenue growth Applying a 25% ebit margin to the 37B KRW in orders so far gets us 9.25B KRW in ebit in Q2 alone You are getting all of this profitability and growth for only ~160B KRW market cap with the current share price at 36,000 KRW If we take Q1 ebit 5B KRW + Q2 ebit 9.25B KRW + Assume Q3 and Q4 ebit stays the same as Q2 (18.5B KRW) = 33B KRW EBIT 5x forward ebit multiple for a company whose Q2 revenue growth was 100% QoQ Is this cycle peak profitability or durable revenue/earnings growth? Everything related to memory is cyclical we know this. But is the market pricing the reality of the current cycle. I don't think so. CXMT CXMT recently filed for an IPO in December last year, seeking to raise 29.5 billion yuan (approx. $4.2 billion) CXMT is expanding its Shanghai fab, aiming for two to three times the capacity of Hefei headquarters. With equipment installation in late 2026 and mass production starting in 2027. Longer term, the company plans to expand capacity to 300,000 wafers/month by year-end and beyond 400,000 thereafter using IPO proceeds. + YMTC investing in expanding NAND manufacturing Its Wuhan campus plans include a new facility where an estimated half of production will target DRAM. YMTC is fast-tracking the Wuhan Phase III NAND fab, bringing its mass production target forward to the second half of 2026, roughly a year ahead of the original 2027 schedule.  The Phase III project broke ground in September 2025 under an operating entity established with registered capital of CNY 20.72 billion (~US$2.98 billion) Phase III will reach 50,000 wafers/month by 2027 and 100,000 wafers/month at full capacity.  Beyond Phase III, three sources told Reuters the company aims to add two more fabs of equivalent scale. + SK HYNIX planning to double wafer capacity by 2030 Memory Capex is expected to go on for another 3 years minimum. With Mirae at 5x forward ev/ebit they will earn almost the entire market cap in cash flow over the next handful of years. 8x forward ev/ebit = 260B KRW / 4.5 m = 58K/shr 10x forward ev/ebit = 330B KRW / 4.5m = 70K/shr 12x forward ev/ebit = 360B KRW / 4.5m = 80k/shr I AM LONG MIRAE. NFA.

My largest position is $2316.TW, WUS Printed Circuit Co Ltd: an AI PCB maker trading at ~11x look-through P/E and only ~23% of its listed $002463.SZ stake NAV. The stock is ~NT$168. Its listed $002463.SZ stake alone is worth ~NT$733/share gross. That is a ~77% discount to listed stake NAV before giving any value to the Taiwan PCB business. This is not a trapped holdco. They’ve already monetized part of the stake. Why cheap? Holdco discount. China exposure. PCB cyclicality. Low coverage. Monetization risk. No clean “AI” label. But the setup is changing. AI servers need more complex, high-layer PCBs. Capacity is tightening. Q1 net income was up ~141% YoY. $2316.TW hit/closed at limit-up 5 times in May alone. This hits exactly what I look for: Hidden NAV. Downside protection. Positive momentum. AI bottleneck exposure. Inflecting fundamentals. A monetization catalyst. Dirt-cheap Valuation. The market sees a forgotten PCB holdco. I see a discounted tollbooth in the AI supply chain, backed by a listed stake worth over 4x the current share price. Just reaching listed stake NAV implies ~336% upside. That is The Scarcity Trade.

$2316.TW