As the year 2025 is coming to a close, Wall Street is caught between two currents: the increasing skepticism about AI trading, which has been a significant driver of profits this year, and the long-standing seasonal pattern that has historically supported the market in December for nearly a century.
This tension has made investors hesitant about whether to follow the stock price trend or prepare for a market correction.
Crowded trading does not provide easy money.
The Santa Claus rally phenomenon, which encompasses the last five trading days of December and the first two days of January, has resulted in the market achieving positive returns 79% of the time since 1929, with an average return of 1.6%. Additionally, in the past eight years, the market has only declined once.
But at the same time, skeptics argue that this pattern has become too well known to be effective anymore. One investor wrote on X, "Seasons tend to work until everyone believes they do, and this is the clearest trade of the year, which is the problem." For this reason, the main reason is simple: the market tends to punish consensus rather than reward it.
Other risky assets besides stocks are also beginning to show weakness, with Bitcoin currently trading at around USD89,460, down 6.9% over the past month, having failed to hold above USD95,000 at the end of November. The current market value of this cryptocurrency is approximately 1.78 trillion USD.
A critical moment for AI.
A more concerning factor is that in the AI sector, which has driven the S&P 500 bull run worth 30 trillion USD over the past three years.
According to a Bloomberg report, signs of skepticism have been continuously increasing, from the recent sell-off of Nvidia stock to Oracle shares dropping after reporting higher-than-expected AI spending, along with changing attitudes toward companies associated with OpenAI. Jim Morrow, CEO of Callodine Capital Management, stated, "We are now in a phase where serious proof is needed. It's a good story, but now everyone needs to decide to put money down to see if the investment returns will be worthwhile."
The cost burden for these companies is very high, with Alphabet, Microsoft, Amazon, and Meta expected to spend over 400 billion USD on data centers in the next 12 months, while the depreciation of these companies combined will triple from about 10 billion USD at the end of 2023 to 30 billion USD by the end of 2026.
A Teneo survey cited by the Wall Street Journal found that there are currently fewer than half of AI projects generating returns greater than costs. Nevertheless, 68% of CEOs plan to increase AI spending in 2026. The survey indicates that marketing and customer service are the areas of AI that yield the highest productivity, while applications in security, law, and human resources are still lagging behind.
Moreover, there is still a gap in expectations, with 53% of institutional investors expecting to see returns within six months, while 84% of CEOs of large companies believe it will take longer.
A case for optimism.
However, comparing it to the dot-com crisis may be exaggerated. The Nasdaq 100 is currently trading at 26 times expected earnings, which is still below levels of more than 80 times during the 2000 bubble. Nvidia, Alphabet, and Microsoft are all trading at margins of less than 30 times.
And looking back at history, the bull side has had better odds. According to data from the financial newsletter The Kobeissi Letter, the last two weeks of December are considered the best time for stocks in the past 75 years, with the S&P 500 possibly reaching 7,000 points by the end of this year.
In the short term, seasonal strength and fear of missing out may continue to support the market, but as we enter 2026, whether investments in AI will provide real returns will be a crucial variable determining the market direction.
