How To ACTUALLY Trade World War 3
Speakers
Thread Guy — Host
— crypto podcast host (aka Red / The Red Guy)
Summary
- The guest manages over $1 billion with a focus on US equities and semiconductors; currently holds a high cash position (~45%) due to geopolitical risks and market valuation.
- The "WW3" narrative is overstated; the US administration's primary goal is keeping oil under $90 to manage inflation and interest rates.
- A "SAS Apocalypse" is real for low-complexity software (Adobe, TurboTax), but high-complexity enterprise software (Synopsis, SAP) has a deepening moat.
- The semiconductor cycle is shifting from simple compute to power efficiency (800V architecture) and advanced packaging, favoring niche European equipment makers.
Trade Ideas
The US administration needs oil prices low to manage inflation and interest rates. If oil (WTI) breaks above $90, it forces interest rates higher and removes the Fed's ability to support the economy. This would be the signal to de-risk equities. Watch. As long as oil is <$90, the "market put" exists. If >$90, short equities. Geopolitical escalation in Iran spirals out of control.
Memory chips are a commodity. Margins rise and fall for all players (Hynix, Samsung, Micron) simultaneously. Hynix is trading at ~4x earnings. When cyclical commodity stocks trade at low single-digit P/E, it usually signals the peak of the cycle. The market is pricing in a future earnings collapse. Avoid. There is a 60% chance the cycle blows up within a year. This turns into a "super cycle" where memory is no longer treated as a commodity (20% probability).
Nvidia is vertically integrating the entire rack (chips, networking via Mellanox, software), while competitors like AMD must cobble together parts from different vendors. This integration allows Nvidia to optimize performance in ways competitors cannot catch. The upcoming GTC conference will likely reveal architectural leaps (Blackwell) that widen the gap. Long. The valuation is reasonable because earnings power is underestimated due to their monopoly on the "full stack" of AI infrastructure. Capital expenditure cuts from hyperscalers (Microsoft, Meta) would crush the stock.
Generative AI can easily replicate the output of "low-end" users (e.g., creating a poster, filling out a simple tax form). Companies like Adobe and Intuit (TurboTax) rely on volume from these basic users. If an LLM can generate a PDF or file taxes for free/cheap, their pricing power and user base collapse. Avoid/Short. These are "bad SAS" companies where the output is easily commoditized by AI. These companies successfully integrate AI into their workflows to retain users.
Synopsis (and Cadence) form a duopoly on EDA software used to design chips. Engineers are trained specifically on this stack and cannot switch. Nvidia invested $2B in them. AI cannot hallucinate complex chip architecture. As chips get harder to scale (Moore's Law slowing), simulation software becomes *more* critical, not less. The acquisition of Ansys (ANSS) secures the physics simulation vertical. Long. It is a "pick and shovel" play on chip complexity that AI cannot disrupt. Valuation is high (~30x earnings), though justified by the moat.
AI server racks are moving from 48V to 800V architecture, requiring massive amounts of power chips (SiC/GaN). Chips also need to be stacked closer together (Hybrid Bonding). Aixtron (AIXXF) has a near-monopoly on the deposition equipment for power chips. BE Semiconductor (BESIY) leads in hybrid bonding packaging. These are the bottlenecks for the next generation of Nvidia racks. Long. These are overlooked "second derivative" plays on the physical constraints of AI scaling (heat and distance). Niche equipment cycles can be volatile; liquidity in OTC/ADR markets.
Apple has been developing its own silicon for years and is launching hardware capable of running AI locally (Edge AI). As cloud inference costs remain high, there will be a push to run AI models locally on devices. Apple is the only hardware player with the silicon efficiency to do this effectively. Long. It acts as a "natural hedge" to Nvidia/Cloud dominance. Lack of perceived AI innovation compared to hyperscalers.
