Okay, not all semiconductor companies, but some.
Note that I write this before some key milestones are reached, such as an AI company going public, so I'm making a prediction here that won't seem like a prediction because I don't plan to officially launch this site publicly until after some of these events happen.
I've long watched the tech industry, and it's all about watching the money. Specifically, the VC money.
Right now the VCs all want Anthropic (and OpenAI) to go public because they'd get their unicorn payback. I've been following some of these investments, and one I know about would produce a 9,233% return at the current valuation (not sure of the number of years here, but probably over 100% annualized).
I can't tell you how excited (and pushy) the VCs get when one of their portfolio companies generates this level of return. Basically, it will validate the entire portfolio at the high return levels they promised investors, meaning any other companies they profit from in the portfolio just make them look brilliant.
Here's the problem as I write this: Anthropic is likely the first of the AI companies to file to sell shares to the public. They currently value their company at about US$2 trillion. Unfortunately, during their dog-and-pony show prior to their filing, I think that they're finding that maybe they can only get US$1.5 trillion valuation by listing. They've filed their S-1 form now, so we can see into their finances to date, and I see many are now saying that they're finding reasons those numbers don't seem to support the current valuation.
While even a US$1.5 trillion valuation will still generate a great return for early investors in the company, the reason for the drop in valuation expectation is troubling: their sales growth rate has declined, and now everyone with too much money in investments is wondering whether or not AI is being oversold.
Anthropic buys servers from Amazon and others. Amazon and others buy chips from Nvidia and others. Nvidia buys tech from ARM and fab from Taiwan Semiconductor. Taiwan Semiconductor buys equipment from ASML. So any hesitation over how much Anthropic is worth will trigger selloff of all those other companies, as they're currently valued based upon lofty AI growth expectations.
The list of companies that fall into this problem is pretty impressive: Amazon, ARM, ASML, Google, Micron, Microsoft, Nvidia, Oracle, SK Hynix, SpaceX, Taiwan Semiconductor, and many more you might not have heard of.
But I'm going to take Taiwan Semiconductor (TSM) as an example here. They are (as I write this) down 17% from the peak they hit earlier this month. Why? Well, as the lack of confidence of how much Anthropic might actually be worth started to circulate, we saw a quiet sell-off of TSM.
However, as far as I can tell, Taiwan Semiconductor not only is at full fab utilization, but has plenty of demand for more fabrication. As you know, I follow the camera companies. They'd love to get more System-on-a-Chip (SOC) done faster, but they're being pushed off the fabs they need for the AI-related chips. Even Apple is feeling this pressure, and they're one of (if not the biggest) Taiwan Semiconductor customer.
I don't see semiconductor company strength going away any time soon, regardless of how much AI does or doesn't take off.
But there's a bigger picture here, too. Semiconductor stocks have always been considered cyclical. Expansion requires a lot of capital, and high tech chip needs come in waves. So when everyone seems to be trying to buy semiconductors to do new things the semiconductor stocks get overvalued as investors pile on, but when the new things they power start to peak in any way, the investors run away from the stocks because they fear that the capital investments won't pay back.
Semiconductor stocks are long-term plays, even though every article you see in the business press will suggest they're short term plays timed correctly. The only trick is to figure out which ones have clear long-term strategies and the ability to finance those.
For a long time, Intel was one of those long-term plays. But then their strategy got all tangled up solely with x86 ego. No chip dominates forever. You don't put all your eggs in one basket. For about six years, Intel was rudderless and uninvestable in my view (disclosure: I held some Intel stock during this period, which turned out to be good news, as recent events basically doubled my value (just over 12% annualized)).
Finally, my prediction: Anthropic won't go out at a US$2 trillion valuation, but something 20-25% less than that. This will trigger the dominoes: the server providers will go down, the chip makers will go down, the semiconductor makers will go down, the semiconductor equipment makers will go down.
If you're paying attention and know the tech industry well, that will be the time to buy. However, remember that the rest of the investors basically create cycles. It'll be a while before you get a reasonable return by buying on the downturn.
Disclaimer: This article is for informational and educational purposes only and does not constitute financial, investment, tax, or legal advice. The author is not a licensed financial advisor, broker, or registered investment professional. Any figures, returns, or examples are illustrative and based on information believed to be accurate at the time of writing, but accuracy is not guaranteed. Past performance does not guarantee future results, and all investments carry risk, including the possible loss of principal. Readers should do their own research and consult a qualified financial professional before making any investment decisions. The author and publisher accept no liability for any loss or damage arising from reliance on this content.