On November 5th, the day after the results of the U.S. midterm elections were announced, the NASDAQ Composite Index fell by 5.52%. The NASDAQ 100 Index declined by 6.36%, making it the worst day since March 2020. The Philadelphia Semiconductor Index plummeted by 12.3%, achieving its worst daily performance in over six and a half years. The Volatility Index VIX increased by nearly 70% within a single day. The following day, the Korean KOSPI index opened down by almost 9%, triggering a trading suspension.
One week later, NVIDIA’s founder Jensen Huang resigned from his position.
Many people connect these two events together and reach the simple conclusion that the AI bubble has burst. But I don’t think so.
My judgment is that this "storm" doesn’t represent the rejection of AI, but rather a reallocation of power within the realm of AI.
Demand continues to exist, and technology keeps advancing. But who can decide how AI should be developed, where it should be developed, to whom it should be sold, and who should benefit from its profits? The answers to these questions are changing.
What the market is doing is revaluating these changes.
The storm had already begun early on.
The first impression that needs to be corrected is that the storm didn’t start with the elections at all.
The Philadelphia Semiconductor Index and Goldman Sachs’ AI-related portfolio both reached their peak on June 22. By early November, both assets had declined by approximately 36% and 32%, respectively. The NASDAQ index, on the other hand, reached its peak on October 6. In other words, the revaluation of AI-related assets occurred more than four months before voters made their decisions.
The cracks originated from several directions. In September, the Federal Reserve raised interest rates by 25 basis points. On October 28th, there was no action taken, but two members voted in favor of continuing with the rate hikes. At the end of October, Intel provided weak business forecasts, Amazon’s financial results fell short of expectations, and the semiconductor sector experienced a significant decline. The NASDAQ index fell by 2.86% in a single day. Meanwhile, several technology companies began to restrict the use of AI technologies by their employees. Within a week, these companies issued bonds totaling $57.5 billion.
This point is crucial. Over the past two years, the narrative surrounding AI investments has been that “big companies have so much cash that they can spend it endlessly.” But now, as these companies start to limit their internal usage and even borrow money to build data centers, the market naturally turns to a question that wasn’t often asked before: Who will actually pay for all these investments?
A concentrated storm
The second impression that needs to be corrected is that this isn’t a systemic crisis at all.
As of mid-November, the S&P 500 index is still down by only about 6% from its highest point. Seven industries remain above their October highs. The VIX index has also declined from 28.18 on November 5 to 18.70. Data on employment levels, the ISM manufacturing index, and GDP for the third quarter are all very positive.
So, what’s being damaged isn’t the American economy, nor the entire stock market. Instead, it’s a very specific aspect of the economy: AI infrastructure, which is driven by capital expenditures and liquidity. Chips, data centers, emerging cloud service providers, and all assets that believe “AI development will accelerate indefinitely” are all included in this category.
This distinction is very important. In the case of systemic storms, the economy will experience recession, and credit conditions will deteriorate. This requires a different approach to dealing with the situation. In contrast, when there are concentrated storms, it means that there are problems with the pricing logic of certain types of assets. However, the economy itself may still be operating at high levels, or even so hot that interest rates need to be raised. Currently, we are in the latter case.
Continue reading the full article
Sign in as a member or explore membership options.