TIP800: Navigating an AI-Driven Market w/ François Rochon

Episode Summary

Executive Summary: Francois Roshan, founder of Giverni Capital, discusses his 2025 annual letter, reflecting on a tough year where Constellation Software fell 26%, and he sold CarMax and Pfizer after their theses broke. He views AI as a revolution on par with the early internet, but warns of circular investment dynamics in AI infrastructure that complicate valuations. He remains bullish on Alphabet and Meta, seeing their massive CapEx as defensive and growth-oriented, respectively. He argues Constellation Software is cheap post-selloff due to AI fears, praising former CEO Mark Leonard's altruism. He emphasizes rationality, humility, and patience as essential investor traits.

Main Topics: AI as a Revolution and Its Investment Complexities (Priority: 5/5): Francois compares AI to the early internet, noting rapid LLM advancement and shifting leadership. He highlights a circular dynamic where NVIDIA invests in OpenAI, which contracts Oracle, which buys NVIDIA chips, making profit recurrence uncertain and valuation tricky. Alphabet and Meta's AI CapEx: Defensive vs. Offensive (Priority: 5/5): Alphabet's $180B and Meta's $100B+ CapEx are seen as defensive for Alphabet (protecting search from ChatGPT) and offensive for Meta (improving ad targeting). Both have strong moats due to free products and massive user bases. Constellation Software's Drawdown and AI Threat (Priority: 5/5): The stock fell over 50% due to CEO Mark Leonard's health departure and AI fears. Francois argues the business is intact, AI won't easily replace niche software, and the new CEO Mark Miller is well-qualified. He sees the stock as cheap at 18x earnings with 20% growth. Lessons from Mistakes: CarMax and Pfizer Sales (Priority: 4/5): CarMax's moat weakened due to competition from Carvana and AutoNation, leading to permanent margin compression. Pfizer was sold due to CEO departure and debt concerns, despite cheap valuation. Francois emphasizes discipline on debt and recognizing broken theses. Three Essential Investor Qualities: Rationality, Humility, Patience (Priority: 4/5): Francois stresses rationality to see reality, humility to learn from mistakes, and patience to wait for rewards. He cites Charlie Munger and Philip Carret as exemplars, noting that patience is often disguised genius in investing. Market Valuation and Future Expectations (Priority: 4/5): The S&P 500's 14.8% return over the last decade was driven by exceptional growth from mega-cap tech and multiple expansion. Francois expects lower future returns as size becomes an anchor, but sees many individual opportunities at reasonable valuations. New Positions: LVMH and Universal Music Group (Priority: 3/5): Francois entered LVMH after a 20-year watch, citing strong brands and cyclical recovery potential. He acknowledges China exposure as a long-term positive despite current headwinds. Universal Music Group was added but discussed less.

Key Arguments: AI is a revolution as important as the early internet, but its rapid evolution makes predicting winners and valuing related companies very difficult. The circular investment dynamic in AI (NVIDIA invests in OpenAI, which contracts Oracle, which buys NVIDIA chips) means profits may not be recurring, warranting lower P/E multiples. Alphabet's massive CapEx is primarily defensive to protect its search business from AI threats, while Meta's CapEx has already paid off by improving ad targeting. Constellation Software's selloff is overdone; AI cannot easily replace niche software that handles private data and financial transactions, and the new CEO is well-prepared. CarMax's moat eroded due to increased competition from Carvana and AutoNation, leading to permanent margin compression, justifying the sale after 18 years. Pfizer was sold due to CEO departure and debt levels near Francois's threshold of 5x net income, despite the stock being cheap at 7-8x earnings. The three essential qualities for investing success are rationality (seeing reality), humility (continuous learning), and patience (waiting for rewards). The S&P 500's exceptional past decade is unlikely to repeat due to size constraints on mega-cap growth and elevated P/E ratios.

Data Points: Constellation Software stock decline: 26% - Francois's biggest holding fell 26% in 2025 despite good results. Constellation Software stock drawdown from high: over 50% - The stock fell more than 50% from its all-time high over 6-9 months. Constellation Software earnings growth (2014-2025): 20% per year - Earnings grew 20% annually over 11 years, and stock rose 12-fold. Alphabet projected CapEx 2026: $180 billion - Alphabet's CapEx is projected at $180B for 2026, roughly equal to its annual cash flow. Meta projected CapEx: over $100 billion - Meta's CapEx is projected over $100B, with strong returns from AI ad targeting. S&P 500 return last decade: 14.8% - S&P 500 returned 14.8% annually over the last decade, with 9% earnings growth and multiple expansion. S&P 500 historical earnings growth: 6.7% per year - Long-term historical earnings growth for the S&P 500 is about 6.7% annually. S&P 500 trailing P/E: 25 times - The S&P 500's trailing P/E is around 25, above the historical median of 18. CarMax and Pfizer stock declines in 2025: over 50% - Both stocks fell more than 50% in 2025, leading to their sale. Pfizer P/E at sale: 7-8 times earnings - Pfizer was trading at 7-8x earnings when sold, but debt concerns prompted the sale. LVMH earnings growth potential: 60-70% over 5 years - Francois expects LVMH earnings to grow 60-70% in five years, plus a 2-3% dividend. LVMH tax rate increase in France: 28% to 32% - A temporary tax increase in France raised the rate from 28% to 32% for large companies. Roshan Global Portfolio annual return since 1993: 18.4% net of fees - Francois's portfolio compounded at 18.4% net of fees since 1993, vs S&P 500's 10.8%.

Pivotal Quotes: "I think the greatest quality is patience. And I remember many old stories. I saw an interview with Philip Carret on Wall Street with Louis Rukeyser. I think it was in 1995. ... Louis Rukeyser asked Philip Carret with your 75 years of experience, because I think Philip Carret was 98 at that time. And he said, 'What's your biggest lesson from those 75 years of experience?' And Philip Carret thought about it and said, 'Patience.'" — Francois Roshan: Francois emphasizes patience as the most important investor quality, recalling Philip Carret's wisdom. "I don't remember where I read that, but I think it said that genius in the stock market is really disguised patience." — Francois Roshan: Francois summarizes the essence of successful investing as patience in disguise. "He looks like someone that was in the ZZ Top rock band, you know, with this big beard. But he was like an artist. He was very thoughtful, very sensitive. And, you know, at some point he said that I don't want any salary. So he cut his salary to zero: no bonuses, no stock options. I mean, where do you see that?" — Francois Roshan: Francois describes Mark Leonard's unique altruism and humility as a CEO.

Implications: Investors should be cautious of AI hype and circular investment dynamics, focus on businesses with strong moats and recurring revenues, and cultivate rationality, humility, and patience. The current market offers selective opportunities in beaten-down quality names like Constellation Software, while mega-cap growth may slow.