Citrini on 2024 Election Portfolio Construction And AI “Mania”

Episode Summary

Executive Summary: The conversation centers on Citrini’s evolving thematic investing framework: he remains constructive on AI but is increasingly selective, favoring names with margin of safety and second- or third-order beneficiaries over crowded leaders. He also lays out election-driven baskets tied to tariffs, taxes, defense, reshoring, and crypto, arguing the biggest opportunity is in policy-specific micro bets rather than broad market calls.

Main Topics: AI theme progression and selective positioning (Priority: 5/5): Citrini says AI remains a major long-term theme, but the easy 'picks and shovels' phase has largely played out. He is narrowing exposure toward beneficiaries with better risk/reward, while watching for signs of disruption to leaders like NVIDIA. Election baskets and policy-driven trading (Priority: 5/5): He explains how he structures election-related baskets around expected policy outcomes rather than political ideology, focusing on tariffs, tax policy, defense spending, reshoring/nearshoring, immigration, and crypto. NVIDIA, hyperscalers, and semiconductor competition (Priority: 5/5): The discussion digs into NVIDIA’s extraordinary growth, the risk from hyperscalers developing custom silicon, and the continued cyclical nature of semiconductors despite AI enthusiasm. China AI and domestic semiconductor buildout (Priority: 4/5): Citrini argues China must build an internal AI and semiconductor stack in response to U.S. restrictions, supporting a basket of Chinese chip-related companies and a broader China barbell strategy. Macro risk, recession hedging, and market timing (Priority: 4/5): He says macro remains secondary but recession risk is rising. He prefers to hedge with rates and options rather than make broad directional macro calls until the data becomes clearly bad. Tesla, software, and AI trust/ecosystem winners (Priority: 4/5): He is skeptical of Tesla as a pure AI winner and sees software as vulnerable to AI cannibalization, while favoring ecosystems with trust and distribution such as Apple, Google, and Microsoft.

Key Arguments: AI has moved from the initial 'data center buildout' phase to a more selective phase where winners depend on who captures real usage and ecosystem trust. NVIDIA is not a classic 'bubble' so much as a mania; earnings growth is real, but the stock is increasingly vulnerable to disruption from custom silicon and recession. The best election trades are not broad 'Trump vs. Biden' market bets, but specific policy exposures such as defense, tariffs, tax rates, reshoring, and crypto regulation. Trade policy and tax policy are the clearest ways election outcomes can affect individual securities, while the broad market is still driven more by the economy than the president. Trump would likely steepen the yield curve via fiscal expansion, tax cuts, tariffs, and pressure on the Fed to cut front-end rates. China’s chip restrictions create an incentive for domestic semiconductor investment, making local equipment, fabrication, and AI-chip ecosystem names attractive despite execution risk. Software is at risk because AI can commoditize many services; only companies with strong distribution, trust, or ecosystem lock-in may emerge as durable winners. Hedging with rates/puts is justified because the listener’s stock-picking edge may outperform the index, but macro shocks like recession can still inflict large drawdowns.

Data Points: AI basket performance: up ~60% to 70% - Performance since the basket was launched in June 2023 through June 15, 2024 Satrindex performance: up 100% - From May 30, 2023 to June 2, 2024 NVIDIA revenue growth: $7.2B to $26B - Quarterly revenue cited from Q1 FY2024 to Q1 FY2025 NVIDIA revenue growth rate: +260% / 3.6x - Calculated from the revenue increase discussed in the interview AI basket weight: 10% to 20% of book - Approximate allocation Citrini says he keeps in artificial intelligence Security count in portfolio: 300 names - He notes owning 300 securities rather than 500 in the S&P 500 for diversification and drawdown control Initial security count in AI basket: 130 securities - How broad the original AI beneficiary basket was before narrowing NVIDIA cost basis: ~$30 per share post-split - He says his pre-split cost basis was around $250–300, using the 10:1 split to translate Treasury spread: 2s/10s inversion expected to steepen - He predicts a steepening if Trump wins due to front-end rate cuts and inflationary fiscal policy Tariff proposal: 10% broad tariff; 60% on some Chinese goods - Trump tariff policy referenced in the election discussion Corporate tax rate proposal (Biden side): 28% - He cites a Biden proposal to raise the corporate tax rate Stock buyback tax proposal (Biden side): 4% - He mentions a proposal to quadruple the buyback tax ETF example of clean energy vs energy: ICLN vs XLE - Used as an example of how a naive election trade could fail badly Hedging drag on returns: ~6% to 6.5% - He estimates hedges reduced portfolio returns over the past year by this amount VIX level referenced: ~11 to 12 - Used in discussing the cost/appeal of portfolio protection Federal Reserve / policy timing: 'until November' - He says there is little to do macro-wise until after the election timeframe Chinese semicap exports to China: 'parabolic' - He describes Japanese semiconductor-capital-equipment exports to China as having surged sharply TCJA expiration: 2025 - He references that some TCJA provisions expire in 2025

Pivotal Quotes: "We have entered firmly into a dirigiste bipartisan consensus." — Citrini: On both parties favoring active industrial policy, subsidies, and government involvement rather than laissez-faire economics "The most expensive eighths of any move are the first and the last." — Citrini: Explaining why he is cautious about chasing NVIDIA at elevated prices "I am bullish on AI, period." — Citrini: Clarifying that even when he trims winners or hedges, he still believes AI is the dominant long-term technology trend

Implications: Listeners should expect more policy-driven stock dispersion around the election, continued AI leadership but with rising selectivity, and higher importance of hedging. Broad index bets may be less effective than targeted baskets tied to tariffs, taxes, reshoring, and semiconductor supply chains.