Is the AI Boom a Bubble? What It Means for Your Portfolio

Two things happened this week that tell you most of what you need to know about where AI stands as an investment.

On Friday, October 16, Nvidia wrapped up a week in which it shed about $365 billion in market value, with the semiconductor index down 11.4% over the same five sessions. Then on Wednesday, one comment on an earnings call from Elon Musk about SpaceX’s computing plans had Nvidia, Intel and SanDisk all bouncing in premarket trading, though the gains faded after the open. On the same day, hedge fund billionaire Ken Griffin said AI is “almost certainly” a bubble, while adding that he isn’t betting against it over the long run.

So which is it, a bubble or a boom? The honest answer is that smart people disagree, and you don’t have to settle that debate to make sensible decisions with your own money. What you do need is to understand what’s actually moving, and how much of your portfolio is riding on it without you realizing.

What Actually Went Wrong This Month #

The interesting part is that demand for Nvidia’s chips wasn’t the problem. By most accounts it’s still strong. The trouble was financing.

Building AI data centers costs a staggering amount of money, and a lot of it is borrowed. When government bond yields rose and corporate credit spreads widened this month, borrowing got more expensive, and the market started asking whether the companies doing the building can keep paying for it.

The clearest example was Core Weave, a cloud provider that Nvidia backs (Nvidia owns about 11% of it). Core Weave lost nearly a quarter of its market value in a week, and the yield on one of its bonds jumped above 15%, up from roughly 9.6%. One analyst estimated the company carries about $30 billion in net debt and $36 billion in lease commitments, so even a four point rise in its financing costs would add close to half a billion dollars a year in interest.

There were other warning lights too. The Financial Times reported that at least 47 major US data center projects have been stalled or cancelled this year because of financing, power and planning constraints.

One analyst summed it up as “a financing cycle more than a demand cycle,” which suggests delays and consolidation rather than the whole thing collapsing. It’s also worth noting that the stress is concentrated among heavily indebted operators. Microsoft, Alphabet and Meta are in far better financial shape.

The Bigger Issue: How Much of the Market Is Betting on This #

Here’s the part that matters for ordinary investors, and it has nothing to do with whether you own Nvidia directly.

The ten largest stocks in the S&P 500 now account for roughly 39% of the index’s value. Most of them are tied to AI spending in one way or another, either as the companies building it or the ones selling the chips and cloud capacity behind it. That level of concentration is the highest since 1965, and it’s climbed from about 26% in late 2022. For comparison, the top ten were also around 26% at the height of the dot com bubble.

Valuations are stretched as well. The Shiller CAPE ratio, which compares prices to long run average earnings, is above 40. The dot com peak was around 44.

If you own an S&P 500 index fund, which is what most 401(k)s and IRAs hold in some form, you’re not as diversified as the “500” in the name implies. A big chunk of your money is effectively riding on a handful of companies whose fortunes depend on the same spending cycle. When they stumble together, the index stumbles with them.

For context on what that can look like, the S&P 500 lost nearly half its value in the two years after the dot com collapse. No two bear markets repeat exactly, and nothing says this ends the same way. But that’s the scale of downside that concentrated, richly valued markets have produced before.

The Case That It’s Not a Bubble #

It’s fair to give the other side its due. Analysts at DBS argue that Nvidia’s valuation, measured against its earnings, doesn’t look like a bubble the way the dot com darlings did, many of which had little or no profit. Nvidia is making enormous amounts of money right now.

The skeptics’ counterpoint is about what comes next. Goldman Sachs estimates the big hyper scalers could spend around $1.2 trillion combined in 2027, and has said they need at least $300 billion a year in AI revenue just to break even. Nobody knows yet whether the revenue will show up on that schedule. That uncertainty is the whole debate in one sentence.

And a note of caution about this week’s rebound: Musk’s comments were management targets for computing capacity, not signed purchase orders. Headlines like that can move prices for a day without changing the underlying picture.

What You Can Actually Do About It #

You don’t have to pick a side. A few practical steps hold up whether AI turns out to be the next internet or the next dot com:

Check what you actually own. Open your 401(k) or brokerage account and look at your fund holdings. If you hold an S&P 500 fund, a total market fund, a tech heavy ETF and a few individual tech stocks, you may be stacking the same bet four times. Most fund companies show the top ten holdings on the fund’s page, which takes about two minutes to check.

Diversify on purpose. Funds that equal weight the index, or that tilt toward small cap, international, value, or bonds, own different things than a cap weighted S&P 500 fund. That doesn’t mean abandoning index funds. It means being deliberate about whether a single theme drives most of your returns.

Keep investing on a schedule. If you contribute every paycheck, you’re already doing the thing that works best in uncertain markets. Steady buying removes the pressure of guessing the top. And if you want a way to gauge how nervous the market is at any moment, our explainer on the VIX fear gauge shows how to read it.

Match your risk to your timeline. Someone with 30 years until retirement can sit through a 40% drawdown. Someone retiring in three years can’t. If your date is close, a concentrated market is a reason to look harder at your bond and cash allocation, not a reason to panic.

Be wary of chasing. Sharp rebounds on a single headline are exactly when people buy near the top. Decide your allocation when you’re calm, not on the day a stock jumps 8%.

The Bottom Line #

AI may well be one of the most important technologies of our lifetimes and also be priced for more than it delivers in the near term. Both can be true at once, and history has plenty of examples. The railroads, the internet and electrification all changed the world, and plenty of investors who bought at the peak of each still lost money for years.

You don’t need to predict which version of the future shows up. You need to know how much of your money depends on the answer, and to make sure that number is one you can live with.