The discussion begins with how Big Tech is financing enormous AI infrastructure projects through leases, special-purpose entities, guarantees, and other arrangements that keep liabilities away from traditional balance sheets. This makes the scale of the bet—and the potential losses—harder for investors to evaluate, especially if demand or technical progress falls short. The hosts argue that Wall Street’s enthusiasm has discouraged deeper scrutiny, while the industry’s financial stability depends on unusually strong execution and sustained growth. They then assess the contrast between AI’s transformative promise and the fragility of its current business model. Meta and Google may be better positioned because of their established businesses, but the broader ecosystem remains exposed to disappointing results. Anthropic’s fast-growing revenue and OpenAI’s mounting losses, competitive pressure, and executive departures illustrate how unstable even the leading labs remain. Despite their scale and valuations, they still operate much like startups. Finally, travel is presented as a useful real-world test for AI because it combines research, changing constraints, judgment, and complex logistics.