The tone of AI stock behavior has changed in important ways this spring.
Until March, they were clearly undervalued, and were quietly and somewhat patiently being accumulated by the minority of investors who realized the significance of AI. Recent buying has been less patient, occasionally mildly panicky, and indicates that awareness of AI is steadily spreading toward mainstream investors.
In a normal year, closing the Strait of Hormuz would impact the stock market much more than any other problem. But for the past few weeks AI has clearly been having a bigger impact on the market. Almost big enough for markets to forget about Hormuz. It’s weird that rising oil prices haven’t much hurt non-AI stocks.
Some examples of the new attitude:
MU is up 150% since the March lows.
NBIS is up 125% since the March lows.
AXTI’s doubling time has shrunk from about 9.5 weeks to 5 weeks. It benefits from an indium phosphide shortage that’s expected to become some sort of constraint on datacenter growth. I originally bought a few shares because they were trading below book value and had something to do with datacenters. Now it’s up 70x, and I’m confused about how to evaluate it.
TRT rose 134% in 3 trading days on no apparent news. I guess people started noticing that it is a semiconductor company, and AI is lifting that whole industry. This reminds me somewhat of patterns that were common in February 2020. The recent magnitude and frequency of such patterns still seems less than half of what it was in February 2020. I expect the pattern to become more frequent before the AI bubble’s main peak.
MRAM is up 408% from the March lows, on what looked to me like rather ordinary good news. I’m unclear on how much of this is AI-driven. Maybe its products are needed for Musk’s datacenters in space? Or maybe general enthusiasm about semiconductors? I don’t see any news that would explain more than a 40% rally, so I’m just guessing.
This feels reminiscent of the later stages of the dot-com bubble – the equivalent of somewhere between late 1998 and the fall of 1999. My intuition says we’re at least two months from a major top, and I wouldn’t be surprised if the bubble continues to expand for another two years.
I switched to being a net seller of AI stocks in April, and I’ve been selling steadily but not aggressively – enough to get me slightly below 100% net long (I’m normally more like 130% net long), but my portfolio still has a high beta, maybe 5?
Solar stocks are currently my favorites for new purchases. AI is almost certain to increase demand for solar energy, yet investors are not treating solar as part of the AI trade.
There’s still a fair amount of room for more investors to realize that AI is important. AI will have a bigger impact than the internet did, so we should expect a bigger bubble. Few AI stocks look overvalued now, but if the recent rally continues, I’ll start treating many of them as overvalued this summer. Keep in mind that the recent rally has been unusual enough that we shouldn’t predict a high probability of it continuing. I’ve been selling because the risk/reward ratio has deteriorated a good deal over the past month, and I’m worried about how uncertain the world has become.
Keep in mind that the best parts of the dot-com boom included significant corrections. E.g. Amazon dropped 45% in August to September of 1998, and then dropped 51.5% (from a much higher peak) in January to February of 1999. Possibly the AI stock declines in reaction to the DeepSeek moment and the Liberation Day tariffs were enough to satisfy the expectations of similar crashes in AI stocks, but it seems more likely that we’ll get between 1 and 3 more corrections of that magnitude before a major top.
I expect to see important political implications of this widespread awareness of AI. Feelings of urgency about AI are becoming somewhat widespread. The surge in AI stocks will contribute to that urgency, but AI will generate other types of news, such as the Mythos security issues, that contribute more directly to mainstream concerns.
I expect something like an AI pause to become an important political issue in 2027. Prediction markets are still skeptical of a pause. I’ve been cautiously bidding up prices on these markets:
Which solar stocks, eg CSIQ?
Yes, which solar and AI stocks are you buying this week? I completely missed the boat on AI stocks–not because I lacked the “situational awareness” but, well, I didn’t want to be heavily invested in such a dangerous technology so I only dipped my toe in… now that everyone and their dog is invested I see the futility of divestment. But still I couldn’t invest because I locked up too much money in stocks that lost money and I was waiting for their price to go near my cost basis before selling. Then Hormuz happened and I thought there was room for the market to go lower. Now I feel well and truly f**ked unless somebody can give me some names that haven’t reached the moon yet