tay kim

Meta growing at 33% with unbelievable profits. Core advertising more durable than Google search (which is ad-loaded and losing traffic to AI chatbots). WhatsApp digital assistant (OpenClaw) shows path to consumer AI agent. Stock rerates higher if AI breakthrough or spending pullback.

TSMC Q1 earnings tone changed - capex will be much higher over next 3 years vs last 3. Jensen confident supply issues (memory, wafers, optical) resolve in 2-3 years. TSMC wafer capacity will be dramatically higher, enabling Nvidia to meet trillion-dollar demand.

Nvidia trades at below-market multiple despite 80% revenue growth because skeptics fear peak is near; however, HBM memory demand (25x more per GPU, 25x more GPUs = 625x revenue potential), supply chain lock-up (memory, wafers, optical), and impending massive stock buybacks will drive rerating. TSMC capex expansion removes supply ceiling.

Same HBM thesis as SK Hynix - only three companies (SK Hynix, Samsung, Micron) can produce high-bandwidth memory. Structural supply deficit for 2-3 years with triple-digit revenue growth at single-digit multiples.

Only three companies can make HBM memory. AI accelerators need 25x more memory per GPU and 25x more GPUs in two years (per Michael Dell). Memory companies cut capacity in 2022 and take 3-4 years to expand, creating mega pricing power. Koreans liquidating insurance to buy SK Hynix.

SpaceX IPO prospectus imminent, Goldman lead left. Founders Fund/Valor $60B+ gains, Sequoia $20B+. But valuation of 1.5-1.75T on Starlink revenue implies extreme multiple; even D1 Capital investor appeared uncomfortable with price. Less upside for retail/passive indexes at these levels.

Google non-existent in coding agents (fastest growing enterprise market). Anti-gravity nowhere to be found. By investing heavily in Anthropic, Google risks repeating Yahoo/Google history where the partner becomes the power player. Gemini losing market share to Claude Code and Codex.