Steve H. 3.0 9 ideas

Steve notes a massive divergence where Software ETFs (IGV) are crashing while Industrial ETFs (VIS) are rising. He specifically highlights Hyundai Heavy Industries (shipbuilding) and Caterpillar (heavy machinery). The US has lost its shipbuilding capacity but needs to patrol waters against China (geopolitical friction). South Korea (Hyundai) retains this capacity. Furthermore, AI data centers require physical construction, benefiting heavy machinery (CAT). LONG. This is the core "Bits to Atoms" trade. Physical infrastructure is the bottleneck for the digital future. Global recession slowing down capital expenditure on infrastructure.

Steve highlights Rheinmetall (German tank manufacturer) as a proxy for European defense. He notes the stock rose on the Ukraine war but has further to go due to US isolationism. The US is signaling a return to the "Monroe Doctrine" (focusing on the Americas), meaning Europe can no longer rely on the US for protection. Europe must rebuild its own military industrial base immediately. LONG. Structural demand for European defense assets regardless of short-term peace talks, due to the shift in US foreign policy. Sudden geopolitical de-escalation or peace treaties reducing immediate defense spending urgency.

AI data centers require massive amounts of electricity. Transformers (high to low voltage) are sold out for years. Steve explicitly names GE (nuclear arm), Schneider, and ABB. You cannot have AI without power. The grid is constrained. Companies that manufacture the physical electrical infrastructure (transformers) and generation (nuclear) have pricing power due to supply shortages. LONG. This is a "pick and shovel" play on the energy crisis created by AI compute demand. Regulatory hurdles for nuclear; faster-than-expected efficiency gains in chips reducing power needs.

Steve points out that memory and storage stocks (Micron, SK Hynix) have gone vertical. He references the rising rental price of Nvidia H100s as proof of compute demand. More compute requires more memory. These are cyclical commodities that are currently in a "super cycle" due to the step-function increase in demand from AI agents (which consume far more inference compute than humans). LONG. A direct hardware play on AI scaling laws. Cyclical downturn in memory prices if AI capex slows down.

Steve argues that Software (SaaS) is facing "Knightian Uncertainty." AI allows companies to "vibe code" their own internal tools, threatening the pricing power and moats of enterprise SaaS models. SaaS valuations were based on high growth and infinite margins. If AI reduces the need for seat-based subscriptions or allows cheaper internal alternatives, the premium valuations (PE ratios) must compress significantly. AVOID. Until the "winners" who successfully integrate AI (like Databricks) are separated from the "losers" (generic reporting software), the sector is a value trap. AI adoption actually accelerates SaaS usage rather than replacing it.