Bullish and bearish opinions expressed in this episode, paired with supporting transcript quotes. The quote confirms what was said—not whether the opinion is correct.
Bullish
$NVDA— NVIDIA maintains dominant position with best hardware engineering, strong relationships across supply chain, and ability to quickly match any competitive innovations. Could sell $2-3T of GPUs if not supply constrained by TSMC.
$ANTHROPIC— Added $11B ARR in one month (unprecedented in capitalism history), dramatically lower cost per token than OpenAI, trading at attractive valuation relative to unconstrained revenue potential of $100-200B if compute available.
$TSM— TSMC's capacity decisions are the single most important variable preventing an AI bubble. Their disciplined supply management and technological lead over Intel/Samsung positions them as critical gatekeeper.
$SPACEX— Uniquely positioned to solve long-term power constraints through orbital compute (racks in space), operates world's largest satellite fleet with cooling expertise, and TerraFab joint venture will attract top semiconductor talent globally.
$AMZN— Strong position with Trainium chip performing better than perceived, internal models improving, robotics driving P&L efficiencies, and deep engagement with startup ecosystem provides talent advantage.
$META— Successfully transformed into AI-first company internally under Zuckerberg, Muse model near Pareto frontier, paid up for compute capacity at right time, better positioned with improving rate of change.
$GOOGL— Largest compute infrastructure and installed base, massive valuable data assets including YouTube, GCP accelerating. Despite losing TPU advantage, scale and data position them well long-term.
$DRAM— Structural shortage with cross-sectional valuation anomaly - DRAM trading at mid-single digit multiples while semi cap equipment at 40x. Memory business models improving with HBM, gap cannot persist.
$CEREBRAS— Did something fundamentally different and hard with wafer-scale computing, working on optical wafer integration to solve IO limitations, architectural decisions create defensible moat against fast followers.
$POWER— Capitalism solving Watts shortage by 2027-28 as new energy sources come online, companies ramping turbine capacity, orbital compute provides long-term solution, terrestrial data centers remain valuable.
$MSFT— Satya making courageous decision to use compute internally rather than serving OpenAI, positioning for world where frontier models not API accessible, risky but right long-term strategic choice.
Bearish
$INTC— History shows Intel and Samsung lack discipline - will eventually break on capacity expansion causing oversupply. Intel made self-inflicted wounds allowing TSMC to catch up, structural execution issues persist.
$OPENAI— Burned 80% more capital than Anthropic to reach similar revenue scale, dramatically higher cost per token, capital efficiency concerns relative to competition.
$SAAS— Private credit struggling with SaaS loans that need further markdowns, debt-laden companies cannot adapt to AI-driven margin structure changes, structural challenges for legacy SaaS business models.