In 1996, Alan Greenspan, then chairman of the US Federal Reserve, delivered a now-famous speech to the American Enterprise Institute, warning against the risk of "irrational exuberance." He denounced the stock market euphoria fueled by the internet revolution and the temptation to justify any valuation in the name of technological progress. Nearly thirty years later, the parallel with the enthusiasm surrounding artificial intelligence (AI) is striking.
The figures are indeed staggering: nearly $400 billion in AI-related investments this year, and more than $500 billion expected next year. As we mentioned in our last Monday briefing, with such amounts, it's difficult to imagine any short-term disappointment in results. The question of the profitability of these investments, as well as that of potential overinvestment, will only arise later. For now, the results published last week by Amazon, Google, and Microsoft showed that investments in AI infrastructure are already profitable for these digital giants, given the explosion in demand for computing power. The improvement in their profitability is, moreover, one of the main differences compared to the dot-com bubble of the 2000s.
However, Meta's results reminded us that investors are not entirely blind. Because in the game of one-upmanship, at which Mark Zuckerberg excels, the market eventually begins to wonder how these capital expenditures will ultimately be financed. Until now, the mountains of cash held by the GAFAM companies and their abundant cash flows allowed them to self-finance, but when cash flow generation forecasts no longer keep pace with ambitions, the punishment is immediate: a 11% drop in Meta's stock price after the earnings release. Fortunately for the Meta CEO, the credit market is overflowing with liquidity: the group was thus able to easily place $30 billion in new debt following its earnings release, against $125 billion in demand (!).
Last week was also marked by the Fed meeting, which ultimately had a greater impact than expected. It first confirmed the end of Quantitative Tightening (QT) as of December 1, 2025, after more than three and a half years of balance sheet reduction. This decision should help alleviate the tensions observed in the US repo market (see chart of the week). Indeed, since mid-2025, massive Treasury issuances combined with the Fed's balance sheet tightening have drained some of the liquidity from the system, pushing a growing number of players to resort to the central bank's repurchase facility.
In parallel, the 25-basis-point reduction in interest rates, although widely anticipated, met with strong disagreement within the FOMC, with one member arguing for maintaining the status quo. Furthermore, while Jerome Powell considered that "the outlook for employment and inflation had hardly changed since the September meeting," a not surprising position given the lack of economic data due to the government shutdown, he surprised investors by specifying that another rate cut in December was not "a foregone conclusion, far from it." As a result, while the market still valued the probability of further easing in December at over 90%, it is now only estimated at 65%.
This nevertheless resembles a game of deception on the part of Jerome Powell, who is trying to keep as many options open as possible in the current uncertainty, but it is difficult to see, as things stand, what could justify him passing on a rate cut in December. This is not the case for Christine Lagarde, however, who seems to firmly believe in her strategy of playing for time...
