As the market buzzes with discussions about how artificial intelligence will reshape the world, the name of a seemingly unrelated company is frequently mentioned: Peloton. The painful journey of this connected fitness equipment maker, which plummeted from a peak valuation of $50 billion to just around $2 billion, is now being viewed by investors as a mirror reflecting the current AI boom.
From Myth to Joke: What Did Peloton Do Wrong? During the pandemic, Peloton's stock price surged 750% in nine months, betting that the habit of home fitness would become permanently ingrained. However, as the world reopened, demand receded like a tide. The company's revenue has declined for four consecutive years, with the number of connected fitness subscribers dropping 7.6% year-over-year, while the CEO admitted that consumers are shifting their spending towards travel and experiences.
Peloton's mistake was misjudging a temporary surge in demand as a permanent new normal and expanding production capacity frantically based on that assumption. When expectations failed to materialize, the stock price crashed by 94%, serving as the harshest punishment for investment based on demand hypotheses.
The AI Boom: A Story of Déjà Vu? A similar scenario is now playing out in the AI sector. Major tech companies are betting that AI will change the world like the steam engine and are making unprecedented capital expenditures accordingly. Morgan Stanley estimates that the capital expenditure of the five largest tech giants will reach $800 billion by 2026, triple the amount in 2024.
Like Peloton, the core assumption underpinning these expenditures is that demand for AI will maintain rapid growth indefinitely.
Three Warning Signs to Heed First, despite the surge in AI spending, commercial monetization remains in its early stages. Gartner predicts corporate AI spending will reach $2.5 trillion this year, but PwC points out that nearly three-quarters of AI's economic value is captured by just one-fifth of listed companies, indicating that a significant portion of investments has yet to yield returns.
Second, tech companies are supporting their AI build-out through massive debt issuance. The free cash flow of Amazon and Meta is already approaching negative territory, with new capital expenditures almost entirely funded by issuing debt. Morgan Stanley expects issuance in the US investment-grade bond market to reach $2.25 trillion this year, with the tech sector contributing 18% of the supply.
Third, the debt market is showing signs of fatigue. While Meta's $25 billion bond issuance attracted $96 billion in orders, this demand was significantly lower than the $125 billion seen in a comparable offering last year. Furthermore, a $38 billion loan for an Oracle data center took six months and still hasn't been fully placed.
What Will Determine the Outcome? Peloton's collapse teaches us that when an investment cycle driven by narrative fails to deliver on its demand promises, the reversal can be swift and brutal. Whether AI will follow a similar path depends on the answers to the following questions: When will the return on AI investment materialize? Can demand growth outpace the speed of debt expansion? And when will credit markets tighten, or even close?
Goldman Sachs warns that the investment-grade bond market is increasingly resembling the stock market, with high concentration in financing AI infrastructure, yet fixed income offers no upside potential. Morgan Stanley succinctly points out that the credit market is financing the AI build-out. Once the credit market closes, the AI super-cycle will come to an end.
Peloton shows us that when financing costs rise and demand assumptions are shaken, any bubble will burst. AI may not be an exception, unless it can rewrite economic rules faster than home fitness did during the Trump administration.