🧠 AI Infrastructure & Semiconductors

• HBM is breaking the old DRAM cost curve. From 1957 to 2020, DRAM cost per Gb fell by roughly one order of magnitude every five years. AI infra demand and HBM are changing that playbook. Cost deflation is no longer the clean Moore’s Law trade it used to be. 1

• $NVDA still has a monster revenue runway. The bull-case math points to Nvidia exiting 2028 at more than $1.8B in revenue per day. That is hyperscaler-scale monetization, not just chip-cycle upside. 2

• Memory is shifting from boom-bust commodity to structural pricing power. Memory stocks are on pace for over $2T in profit over the next three years: Samsung ~$842B, SK Hynix ~$678B, $MU ~$481B. The key driver is supply-constrained, high-margin products that are hard to manufacture. 3

• $NBIS is trying to catch the next AI cloud leg by scaling from one data center to gigawatt scale AI factories in just two years. The edge is not just capacity. It is rebuilding the platform for the AI era. 4

🚀 Space & Launch Stocks

• $RKLB is getting hit by SpaceX IPO pressure, with the stock down 10% on listing day, briefly breaking below $100, after falling from a $150 high to just above $100. The real question is whether fundamentals actually cracked or whether this is a post-hype shakeout creating a buy-the-dip setup. 5

• Space launch valuations can be detached from reality, but reliable launch is the scarce asset. Out of 142 small-launch startups tracked when Rocket Lab was founded, only two reached reliable orbital launch. That makes scaled reliability the moat, not just TAM slides. 6

🏦 Market Structure & Breadth

• Small-cap financials just closed the week at their highest levels in history. That is a strong breadth tell. Risk appetite is not only hiding in mega-cap tech. 7

🧩 Mega-Cap Tech Re-Rating

• Google is now framed as the highest-conviction AI compounder in U.S. equities. The setup mirrors last year, when the stock was still below $200 and bears kept calling it obsolete and vulnerable to AI disruption. That same stale bear case is now being recycled against Microsoft. 8

• $MSFT valuation math still leaves meaningful upside. Current 2030 consensus EPS is $32. If the multiple rerates back above 30x, the stock can reach $1000. Even if the multiple stays compressed at 20x, the stock still clears $600+. 9

• Several high-quality stocks are near 52-week lows where the stock looks broken but the business may not be: $PLTR, $MSFT, $META, $SOFI, $MELI, $SE, $UBER, $NFLX, $MA, $GRAB. This is a classic dislocation screen, not a pure momentum chase. 10