Dr Jeffrey Funk and I discuss how stock investors tend to focus on how high prices can rise, while bond investors focus on whether they get their principal back.
Jeff explains that lenders now see a real chance that one data centre player, CoreWeave, cannot meet its commitments, so its borrowing costs have shot up to around 15%. Anthropic pays more than 10%, Oracle around 8%, and even cash-rich giants like Nvidia are paying more. He calls this a “death spiral”, where higher rates raise costs, losses mount, lenders see more risk, and rates rise again.
Jeff draws on Robert Shiller’s work on bubbles and says what pops one is a change in the narrative. For years, the AI story has promised limitless productivity, scientific breakthroughs, solving environmental challenges, and even longevity.
In jeff’s view, Sam Altman skipped sceptical engineers and sold that story to CEOs and politicians, who believed it, with dire economic and social consequences that we see today. Jeff thinks the narrative breaks when the weakest players, such as CoreWeave and the smaller “neocloud” firms, can no longer fund themselves and investors rush to sell. That would drag down Nvidia, which Jeff describes as “the Cisco of the AI bubble,” since Cisco fell about 90% after the dot-com crash.
We discuss that the next test for the AI bubble’s future is earnings. Jeff expects Meta, Microsoft, Amazon and Alphabet to show flat revenue growth and that data-centre and AI-lab revenues have been flat for roughly six months. Not to mention, Oracle also reported flat quarterly revenue, which is a serious concern for investors.
Spending has also climbed steadily, so revenue must rise sharply to justify it. Hence, flat revenue is a problem, and falling revenue is even a bigger problem. He also criticises OpenAI and Anthropic for releasing partial, adjusted figures and no audited accounts, while ignoring such costs as share-based compensation.
Jeff sees the same trend in the startup world, where profits used to come within a few years. He found only 24 companies founded since 1975 that reached the top 100 by market value, and nearly all were profitable early. Today, on the other hand, VCs and insiders cash out early, and retail investors absorb the losses.
Beyond the markets, our discussion links the bubble to inequality, higher prices and the unrest now visible in Europe. Jeff argues that business schools and academic economists are unlikely to reassess themselves, because they reward quantifiable, internally consistent research over practical questions about whether a technology is useful. His advice is to follow the basics: who is paying, what the revenues and profits are, and whether the price covers the cost.
We discuss the future after the AI bubble crash, where entrepreneurs will be forced to, or will choose to, bootstrap, find a first paying customer and build from there, which Jeff thinks may work better.
Join the conversation and tell us whether you think the AI bubble will finally burst this year, and what consequences we will all face.
