The AI Issue Is Worse Than I Thought

It seems not a day goes by without something happening in the AI trade. Every time I look down at my phone, it’s the first thing that appears on my Yahoo Finance screen. As you’ll read later in this post, I do think there are parts of the AI trade that make sense and are backed by fundamentals (hence the title “Issue” rather than “Bubble”). Yet, there are some glaring problems the market is ignoring. And one in particular is worse than I thought.

Going Back to the Bubble

In a recent article of mine, I argued that I don’t think the AI trade is a traditional bubble, since it’s supported by large companies rather than a slew of startups. I still hold that view, but the bubble-like behavior has become more obvious.

The recent funding round of Anthropic is a great example, where its valuation crossed that of Walmart. For a quick comparison, Walmart’s revenue is over $700B, whereas Anthropic’s is $4.5B. Beyond that is the recent Cerebras IPO. A company making less than $510MM in revenue is currently being valued at over $50B. Its value was also heavily inflated with a more than 100% jump post-IPO. Tell me that isn’t bubble-like behavior.

AI is becoming more of a buzzword for investors than the opportunity it was meant to be. Now I don’t want to dive too deep into valuations, as I covered that in the article mentioned above. But what I do want to talk about is the investments these companies are making, which are meant to justify these valuations.

The CapEx Issue Is Here to Stay

The CapEx numbers we’re seeing from companies over the buildout of AI infrastructure are staggering. But what many don’t understand is that a large part of that money isn’t building anything. Much of it is maintenance CapEx rather than growth.

A recent Entelligence AI study, which surveyed thousands of companies on how their money is being spent, found that over 80 cents on every dollar spent on AI goes toward debugging, rework, delays, and more. That means only a small part of the investment is creating productivity. AI is meant to have a multiplier effect on the effectiveness of company services, but instead, it requires manual fixes for all the issues it causes. Beyond that, technology is advancing so fast that updates are constantly required, creating more spending that doesn’t directly create production.

And in the face of all this, companies are facing real cash flow concerns, which are driving poor sentiment toward Meta, Amazon, and more. The deeper issue is what this means for these companies in the long run. A lot of the CapEx is spent with the justification of long-term ROI. However, we haven’t seen any so far. Plus, if/when the AI buildout is complete, they’d still have to spend immense sums on maintenance. This would erode their cash over time and their ability to invest elsewhere, creating a long-term structural drag.

I’m seeing companies spend without asking the right questions. And even if they wanted to, the risk of falling behind means they can’t afford to slow down.

OpenAI… Where Do I Begin?

All of this sets the stage for one of the most consequential IPOs we’ve seen in a while. We are set to get a $1T+ IPO from OpenAI, yet the company is wildly unprofitable and losing market share. In my last article about the tech sector, I mentioned that they’re projected to lose $14-17B in cash this year. Since then, estimates have ballooned. Even after cutting memory spending estimates, they’re now projected to burn through $27B.

This is a perfect example of long-term consequences hiding in plain sight. While bleeding cash, ChatGPT’s market share has begun to get eaten up by Claude and Gemini. Combine that with estimates that they’ll need over $200B in extra funding to sustain themselves by 2030, and it’s a recipe for disaster. But that doesn’t begin to account for the CapEx issue this post is about.

Let’s say they become profitable, secure the funding they need, and everything works out. They’d still need billions to cover maintenance, fund further growth, and buildouts. What’s worse is that while Google, Amazon, Meta, and others can absorb these costs with profitability elsewhere, OpenAI can’t. It doesn’t have other services to lean on.

What This Means

None of this means AI will fail. The technology is real and has legitimate uses. The demand for it is also strong, and some companies will take advantage of it. But the market is pricing in a future in which these underlying problems don’t exist.

No company is willing to ask these questions out loud, but I hope the market starts scrutinizing them further. Otherwise, this bubble-like behavior might just turn into a bubble.

This article was inspired by what I read here. Check it out!

Disclaimer:

This blog post is for educational and informational purposes only. It is not financial advice. I am not a licensed financial advisor, and nothing in this post should be interpreted as a recommendation to buy or sell any securities. Trading involves risk, and results are not guaranteed. Past performance is not indicative of future results. Always do your own research and consult with a licensed financial professional before making any investment decisions.