Money for the Rest of Us · J. David Stein

How To Invest During a Bubble

August 20, 2025·21 min·1 clip
Science YouTuber Hank Green tells his 4 million subscribers he's moving money out of the S&P 500 because its top 10 holdings now make up a record 38%.
The premise is simple. David Stein is talking about investing when U.S. stock valuations already look unusually expensive by historical measures. He starts in a personal place. A memory of taking his son Brett to VidCon in Anaheim, California leads to Hank Green, YouTube audiences, and a recent finance video that overlaps with concerns Stein has been tracking. It is a slightly winding path, but it works because the episode is really about how a technology story turns into a market story. Then the numbers show up. Stein says valuations are more than two standard deviations above average, with price to book at 5.4 versus an average of 2.7. That is not a timing tool on its own, and he does not pretend it is. It is a constraint. Investors still have to decide what to do while the market is expensive. AI is the center of the episode. The top five holdings make up close to 30 percent of the market exposure Stein is describing, so the bet is not as spread out as it may look. A lot depends on data centers, productivity gains, and technology spillovers. Meta gives him the concrete example. A Bloomberg article about a rural Louisiana data center lets him look at the cost and financing of the AI buildout. The project is described as a four-million-square-foot data center supercluster using five gigawatts of power, enough to power four million homes for a year. The financing is where the episode gets sharper. Meta is using a special purpose vehicle tied to the data center, not simply paying cash or issuing ordinary debt. Stein describes $29 billion of financing, including $3 billion of equity from Blue Owl and $26 billion of investment-grade bonds handled through PIMCO. Because the debt sits in the vehicle rather than directly on Meta's balance sheet, investors get less visibility into the data center economics. The collateral is not especially comforting either. Stein notes that 60 percent of data center cost is GPUs, and those processing assets have a three-to-five-year life. The tension is clear enough: investors may want exposure to real technological change, but they still have to care about valuation, leverage, disclosure, and asset life. Stein closes carefully. Nobody knows when the bubble ends. Valuations could go higher. The episode is general education, not individualized investment advice.

As heard by us

A measured look at bubble risk, using valuation data and Meta's Louisiana data-center financing as the concrete case.

David Stein starts with a family story about his son Brett at VidCon, then settles into a careful argument about bubble risk, rich valuations, and the AI narrative driving the market.

Read the full review in PlayNext →

Why you'd press play

You want a steady guide to investing during a bubble.

Read the full recommendation in PlayNext →
Listen to the show on