TIP823: From Railroads to AI: The Timeless Patterns Behind Market Bubbles w/ Kyle Grieve

What It Covers

Kyle Grieve examines Ron Insana's bubble framework from *Trend Following*, tracing recurring patterns from 1850s plank road companies through the 1990s tech crash to today's AI buildout. Using Kindleberger's five-stage model and Insana's five-ingredient checklist, the episode provides concrete tools for distinguishing overvalued assets from genuine compounders.

Key Questions Answered

• •Kindleberger's Five Stages: Bubbles follow a consistent sequence: displacement (new technology sparks excitement), over-trading (buyers overwhelm sellers), monetary expansion (cheap credit fuels leverage), revulsion (quiet institutional exit while retail buys), and discredit (sentiment reverses completely). Recognizing which stage an asset occupies helps investors calibrate risk. Isaac Newton added to South Sea Company after a 20% drop and lost nearly everything — revulsion is rarely obvious in real time.
• •Insana's Five Bubble Ingredients: A reliable bubble checklist requires five conditions: a eureka-moment invention, easy money and cheap credit, government largesse such as subsidies or tax incentives, favorable economic conditions like low unemployment and strong GDP, and an external stimulant accelerating demand. When all five align simultaneously, speculative excess becomes probable. Investors can monitor each factor independently using public economic data, policy announcements, and industry news.
• •Terminal Value Reality Check: Bubbles represent time distortion — investors compress decades of future cash flow into today's price. Yahoo at peak required 18 billion customers to justify its valuation when the global population was a fraction of that. For any holding, model what the business must achieve in five years to justify its current price. If the required outcome defies basic math or market size, the valuation has entered speculative territory.
• •Price vs. Intrinsic Value Divergence: When a stock rises faster than its underlying earnings, the gap signals bubble risk. RCA compounded net income at 35% annually from 1923 to 1929, yet its price-to-earnings multiple expanded from 15x to 285x — the business was sound but the valuation was catastrophic. Track multiple expansion separately from earnings growth. If share price rises 7x while EPS doubles, as with InMode in 2021, the excess return is pure narrative, not value.
• •AI Bubble Assessment Using Insana's Framework: Applying the five-ingredient checklist to AI today: eureka moment (confirmed), government largesse via the $53 billion CHIPS Act (confirmed), easy money (partial — AI startups like Thinking Machines raised $2 billion at a $10 billion valuation with no product). However, economic conditions are weaker than peak bubble environments, with US GDP growth at 2.1% in 2025 and unemployment at 4%. Public participation via IPOs remains subdued at $38 billion versus $142 billion in 2021.
• •Three Layers of AI Risk: Separate technology risk, business risk, and valuation risk when evaluating AI investments. Transformative technology does not guarantee investor returns — the US automotive industry started with hundreds of manufacturers and consolidated to three, a survival rate below 1%. Avoid companies funding the AI buildout speculatively; instead target businesses already generating profits from AI capabilities. Use forward earnings multiples: Nvidia's forward PE of roughly 24x sits below the S&P 500's 27x, which does not signal bubble conditions.