The following trades are ordered by conviction level and emphasis as discussed in the interview.
The highest conviction trade for 2026. AI has created a "capability gap" where technology can replace undifferentiated labor (e.g., logo alignment, basic admin), yet companies are still employing humans for these tasks.
The trade targets companies that are currently "people factories" (high headcount, bloated structures) that often trade as "AI Losers." As they inevitably implement AI to reduce headcount and SG&A, their margins will expand significantly, leading to a rerating. This is not about AI capex, but AI efficiency adoption.
Moore's Law has hit a physical "reticle limit"—chips can't get any bigger. The solution is Chiplets: stitching many small chips together. This makes "Advanced Packaging" the critical bottleneck, specifically TSMC's CoWoS capacity which is fully tapped out.
The trade focuses on companies that handle the overflow capacity from TSMC or provide the interconnect tools ("staplers") to bond these chiplets. This theme works regardless of who wins the chip war (Nvidia vs. Custom Silicon) because all high-end chips require advanced packaging.
A derivative of the custom silicon trend. You cannot design a modern chip without EDA (Electronic Design Automation) software. Synopsys holds a near-duopoly in this space.
With the acquisition of Ansys, Synopsys positions itself as the "physics engine" of chip design—simulating heat and warping in 2.5D/3D stacked chips. The stock has been punished due to an Intel contract dispute, offering a buying opportunity in a monopoly-like asset.
Moving closer to the physical "box" of AI. The supply chain relies on specific, commoditized materials where one company may hold a 70-100% market share. When shortages hit, these "boring" chemical stocks can skyrocket.
Key materials include Non-Conductive Film (NCF) for stacking HBM (High Bandwidth Memory), ABF Substrates (insulation layers), and T-Glass (glass that doesn't expand when hot).
AI data centers have insatiable power demands. While solar and nuclear are part of the mix, Natural Gas is the immediate baseload solution ("Machine God needs gas").
Hyperscalers (e.g., Microsoft, Google) are expected to start signing fixed-price contracts with Natural Gas producers to guarantee supply, similar to recent deals with power producers like Vistra. This would allow gas producers to rerate from volatile commodity plays to growth/infrastructure plays.
Industries that were "traumatized" by previous cycles of over-expansion followed by crashes (e.g., Solar, Lithium, Offshore Drilling).
Executives in these sectors are now "once bitten, twice shy." Despite rising demand, they refuse to build new capacity (Discipline). This lack of supply response creates a "super cycle" where prices and margins stay elevated for incumbents.
For AI agents to be truly useful (e.g., "book me an Uber"), latency must drop from ~800ms (cloud roundtrip) to ~200ms. This requires processing on device (Edge AI).
The bottleneck is RAM. Running inference locally requires massive memory. The trade is to long the enablers of edge compute and short the hardware makers who will suffer from skyrocketing RAM component costs.
A capital structure arbitrage. MicroStrategy ($MSTR) has issued preferred securities that offer limited upside (capped yield) but retain significant downside risk if Bitcoin prices fall (dividend suspension).
The Trade: Short the Preferreds (poor risk/reward) and Long Bitcoin (uncapped upside).
World Cup 2026: Hosted in North America. Budget hotels have performed poorly but will
likely sell out completely for the event.
Pick: Choice Hotels ($CHH).
Tax Refunds: Fiscal transfers in Q1 2026 are projected to be 30-50% higher than usual. Consumers will likely spend this windfall on deferred durables like mattresses.