When one sector begins to drag the rest of the market with it, a certain nervous energy permeates trading floors. You can hear it if you walk through any financial district right now: the voices are lowered and people are staring at ticker screens for a little too long. For the better part of three years, semiconductor stocks have been doing something extraordinary, and not everyone is comfortable with how extraordinary it has become.
Recently, chip stocks increased their market value by over $3 trillion over the course of an incredible 17-day period, with the PHLX Semiconductor Index recording an unbroken winning streak. That’s the kind of figure that makes seasoned investors sip their coffee and remain silent. Such actions may represent true structural demand, and there is a strong argument for that. There’s also a chance that something else is going on, something more recognizable and hazardous.
With its share price up nearly 29% in just three months at one point and $46.7 billion in revenue for its most recent quarter—a 56% increase from the same period last year—Nvidia continues to serve as the narrative’s central figure. As you watch that happen, it’s difficult to avoid thinking of other instances in which a single business came to represent the aspirations of an entire market. The zeal seems sincere. In 1999, when everyone was talking about Cisco and Intel and the Nasdaq was doing things no one had ever seen before, it also felt real.
By mid-December 2025, the combined market capitalization of the top ten chip companies was $9.5 trillion, up 181% from $3.4 trillion in late 2023 and 46% from $6.5 trillion just a year earlier. Perception is distorted by those numbers. The distinction between momentum and mania becomes uncomfortable when valuations rise so quickly. The fact that the top three chip stocks alone hold 80% of the entire market capitalization is especially startling; even optimists should be concerned about this concentration.
Even for experts, choosing winners has become extremely challenging due to the industry’s extreme volatility and recent price fluctuations in the chip sector. Both genuine risk and genuine opportunity are created by the combination of volatility and robust long-term demand. The demand is real in and of itself. The development of AI infrastructure is using chips at a rate that was nearly unthinkable just five years ago. Where there used to be farmland, data centers are being built, and the capital expenditures that support them come from actual businesses. However, futures that arrive later than anticipated or in a different way than anticipated can be priced in by markets.
Conversations with investors who have lived through the 1999 internet bubble frequently bring it up. Although most people strongly qualify it and it’s an uncomfortable comparison, it keeps coming up. The demand for AI chips is perceived as being less speculative than dial-up internet revenues, suggesting that the fundamentals are stronger now than they were back then. Regardless of one’s opinion, the move’s speed and scope encourage comparisons.
Global chip sales are predicted by the Semiconductor Industry Association to reach $701 billion by the end of 2025 and to grow at a rate of 26% to $975 billion in 2026. Those numbers are truly astounding. However, the industry has been here before, experiencing prosperous periods that ultimately came to an end due to inventory gluts, geopolitical tensions, or just a buying cycle’s exhaustion. As recently as 2023, the global semiconductor industry’s revenue dropped by 8%, but in 2024 it experienced a sharp increase. This type of whiplash is ingrained in the industry.
Whether the AI wave is a truly new structural reality for semiconductor demand or a particularly well-funded version of cycles the industry has already experienced is still up for debate. The market appears to be stretched, and some analysts are keeping a close eye out for any indications of an impending correction. Investors who are placing bets on sustained growth are presenting a compelling case. They are also the ones who subtly reduce exposure. Both cannot be totally correct, and the rest of the market will most likely notice if one of them proves to be incorrect.
