It has been called by many pundits the biggest speculative event since the late ‘90s Internet stock mania. I’m referring of course to the widely referenced “AI bubble” that has been in play for the better part of the last three years.
But is it truly a “bubble” in the historical sense of the term? The answer to this question is salient for us not only as investors, generally speaking, but also as it concerns at least a couple of the stocks in my Cabot Turnaround Letter portfolio and (presumably) your portfolios as well.
Another reason why I view it as timely that we address the proverbial elephant in the room is that the AI buildout is a big part of my—and many others’—overall investment thesis for the next couple of years. One doesn’t have to be a Mensa member to realize the magnitude of this once-in-a-generation event and its spillover impact on virtually the entirety of the U.S. equity market.
Among its other attributes, the AI buildout is providing a level of interest in stock market participation that hasn’t been seen in probably the last 30 years, with its attendant liquidity proverbially “lifting all boats” across the market.
Indeed, the level of participation we’re now seeing among retail traders/investors was altogether lacking during even the housing bubble years of the 2000s, and certainly during the post-credit crash decade, possibly even rivaling the participation levels of the late ‘90s. As I’ve argued in the past, widespread retail participation carries an extraordinary degree of market dynamism that can’t be underestimated.
Now I’m not going to insult your intelligence by making the hackneyed observation that “this time is different.” There’s obviously a marked degree of speculative froth and “irrational exuberance” in the AI-focused bull market that cannot be denied. As such, the prevailing environment demands a certain level of prudence among safety-minded participants, even as we seek to take advantage of the bull’s remaining prospects.
But is 2026 truly analogous to, say, 1999? I don’t claim to have an objective answer to this, aside from sharing my (admittedly colored) view that today’s market doesn’t exactly “feel” quite as exuberant as it did back then. But what can’t be denied is that the economic backdrop of those heady days was considerably stronger than it is today, with far less inflation, providing an extra “oomph” to the late ‘90s Internet stock bubble—a dynamic that seems to be now missing.
[text_ad]
A much more nuanced view of the AI bubble debate was provided by JPMorgan analyst Susan Bao in a recent Barron’s interview. Asked if we are in an artificial intelligence bubble right now, she emphatically responded in the negative. She pointed out that in the dot-com bubble, “companies laid a lot of unused fiber-optic cables, or dark fiber, in anticipation of demand, and 95% of those stayed dark when the bubble burst” as a result of the obvious over-speculation.
By contrast, today’s AI buildout is attended by what Bao called a “power constraint rather than an advanced-chip constraint.” There are further constraints, she said, related to data-center construction permitting, cooling and water availability, as well as access to specialized labor. These, according to Bao, are the differentiating factors separating 1999 from 2026.
Significantly, Bao further observed that while Internet capacity was built with a heavy reliance on debt during the dot-com bubble era, today’s hyperscalers are using balance-sheet cash, which means today’s AI builders aren’t nearly as leveraged as were their Internet buildout counterparts.
On a related note, the headline of the lead story in a recent edition of the Financial Times caught my eye: “Wall Street turns to complex trades to dodge AI ‘implosions.’” The gist of the article is that institutional investors are restructuring their trades to protect against, or even profit from, potential AI-related stock crashes without abandoning AI exposure altogether.
Aside from using complex tactics involving volatility trades and options strategies, these investors are also hedging their AI sector exposure using a long/short approach to the various categories of the industry, hoping thereby to protect themselves from additional downside risk while at the same time maintaining their investments in the AI buildout.
Essentially, this approach among the major AI investors means that they still believe in AI, but they’re no longer willing to hold it “naked.” Instead, they’re using sophisticated hedges and derivatives to avoid being wiped out if the AI trade suddenly unravels.
Putting aside the obvious concern that this strategy also introduces a potentially dangerous measure of leverage to the overall equation, I view the strategic pivot among the big AI backers as a sign that their support for the sector remains unwavering. And this, I think, means they aren’t likely to categorically dump AI stocks (this year at least), as the dot-com era investors eventually did with Internet stocks in the year 2000, thereby killing the entire bull market.
As such, I’m not presently worried about the AI so-called “bubble.”
[author_ad]
