It’s an AI Bubble! (So Why Are Many Investors Cautious?) Stock Talk Update, Friday April 24, 2026

“It’s an AI bubble.”  If you have been watching many financial media outlets for the last few years, You’ve been hearing that declaration, an opinion, often stated as fact.

The investment team at OHFG has been asked this question for well over 18 months now. Are we reliving the Dotcom Capex boom, internet and mobile phone buildout of 1997-2000, in the AI investment cycle that started late 2023 with the release of Chat GPT or is this something new and different?

But here’s the question I’ve asked for over a year, If it’s such an obvious bubble, and so many pro’s are good at spotting bubbles in advance, why are so many public market equity professional investors still cautious? I mean why aren’t they all-in like 1999?

That disconnect is where the real story is.

We compare the dot-com the stall in the summer of 1999 then surge into 2000 and today’s AI acceleration into 2026.

Think of both as the 7th inning stretch, business momentum was strong, but the stock markets paused for 4-6 months before reaccelerating up.

CAPEX COMPARISON

In 1999, companies were racing to build the internet. Back then it was fiber networks, telecom infrastructure, and eventually the mobile internet. Massive capacity was being built, but it was ahead of real demand. Ahead of Amazon, ahead of Netflix and streaming services, and ahead of most mobile telephone applications we use today. That summer—mid-1999—was the 7th inning stretch.

The real excess in public equity markets came after that. It came in the 6 months during the 4^th qtr 1999 into late March in 1q2000 as spending plowed ahead of end demand. In many ways, today looks eerily similar to that period in late 1999—but it’s fundamentally different. We are three+ years into AI spending. Three years from the launch of ChatGPT being Nov 30, 2022 and usage is exploding from an already high level and growth rate. It is acceleration—driven by real, existing usage and demand for compute.  For systems driven by NVDA and others semiconductor chips. Today, end demand its exploding exponentially thanks to opensource AI programs and agentic agents coming on the scene early February of this year with Open Claw.

AGENTIC AI SHIFT

What changed in the 1q26 that many investors didn’t see until just now?  Just as we exit the 7^th inning stretch? The parabolic AI chat growth in 2023-2025 has recently went exponential in 1q26. This was a tipping point, beyond the VibeCoding moment in 2025. AI is no longer just a tool. It’s becoming agentic—what’s that mean?

AI is now agentic acting, deciding, executing. Systems that act on their own. They run queries. They Make decisions. They give suggestions and Execute tasks—automatically. 24 hours a day, 7 days a week, every day of the year if needed.

This drives real demand and revenue. It’s not “build it and hope”—but rather build it because current capacity is already overloaded and strained. This drives the price of compute up and GPU values up as well not down as Michael Burry postulated in his short thesis about 6 months ago.

 

PRIVATE VS PUBLIC

PRIVATE VS PUBLIC CAPITAL — KEY DIFFERENCE]

Now here is one of the most important differences—and one many investors miss.

In 1999: Capital flowed through public markets, IPOs surged, Speculation spread quickly and Retail investors carried much of the risk.

In 2026: The most important AI companies are still private. OpenAI. Anthropic. Byte dance in China. SpaceX in another area of tech. Their funding comes from: large institutions, Strategic partners and other Big tech platforms not from retail investors, Morgan Stanley and Goldman Sachs.

So the structure is different:

Back in 1999, the risk was largely public. We had an IPO boom and public speculation ran rampant. Today, OpenAI, Anthropic are still private. Much of the risk is still privately held. That doesn’t remove risk. But it changes where it sits.

Risk currently is more contained.

 

RETURNS

Let’s cover the returns during the two-time periods.  Here’s a table of the SP500, NASDAQ Composite, and Sox Indexes from the LTCM Oct, 1998 low into the end of Dotcom 18 months later.

Dot-Com:, S&P 500: +52%, NASDAQ: +184%, SOX: +478%

And here’s the 13 month period from the 10/7/1998 low out to about where we stand today almost 13 months.

S&P 500: +41%, NASDAQ: +112%, SOX: +224%

Followed by our recent 12.5 month returns since the Tarriff tantrum lows 4/8/25.

S&P 500: +42.8%, NASDAQ: +60%, SOX: +176%

In 1999 = everything in the NASDAQ melted up. Pre-revenue companies, eyeball and click companies.  Anything with a sniff of Dot.com in its name.

Today, the markets are strong but selective.  Software stocks have lagged for 12 months as have many of the Mag 7 prior leaders.  The strength has been largely in the picks and shovels as Charles likes to say, semiconductors, optics, infrastructure builders.

ECONOMIC BACKDROP

Now what did the economy look like then and now.

In 1999: Economic Growth was strong; The Federal Reserve was tightening interest rates BUT it was also creating special liquidity in anticipation of potential Y2K disruptions in 1999–2000.

Today: Growth is steady and possibly about to accelerate in 2-4q. Inflation was about to peak at around 3.5% and come down but we got hit by the Iran war so that’s unclear now.

MARKET BEHAVIOR]

As far as financial Markets are concerned, they do look familiar.

In both periods: Leadership was in technology companies, but energy material and infrastructure stocks also did ok.

But here’s the key distinction: In the 1999 stock bubble: Many companies had no earnings, traded at multiples of revenues or eyeballs, and if they did have EPS they traded at 50-100x EPS. See the valuation on CSCO back in the day and compare it to NVDA nowadays. Many Valuations back then were based on hope.

Today: Market leaders are: Highly profitable, generating real cash flow and Seeing earnings accelerate with AI demand. Yes, admittingly, many are now FCF negative given their huge capex spends and moving from asset light to more asset heavy. That is a major change. And that is the risk. the risk today is not: “no earnings.” The risk is paying too much… for very good current earnings and uncertain future earnings and marginal ROIC.

So let’s assume we are in a similar bubble for the sake of the argument.  So where are we right now? A return to our much-shared SP500 overlay, here’s the updated chart. Remember no guarantees.

We look to be exiting the 7th inning stretch where upside remains. Although some risks are rising.

CLOSE

So investors, while this is not 1999, are rhyming a lot. The foundation of the AI capex cycle, profits and existing demand, is much stronger today and valuations are much lower. That’s a good thing and begs the question why so many have been calling bubbles not for a few months, but for a number of years.

Investors, while we are not early in an economic expansion, due to agentic tipping point, we have likely just entered the re-acceleration phase, which should be good for many stocks.

Today, the structure of risk is different. In 1999, it was in public markets and widespread. Today, much of it remains in private markets and concentrated. Maybe the Bubble is in private markets this cycle not the public markets?

Investors, we know history doesn’t exactly repeat, but given humans repeat many past behaviors, and studying behavioral finance, it often rhymes.

Right now, we’re continue seeing echoes of 1999…but with a stronger foundation underneath and just exiting the 7^th inning stretch.

Stay disciplined. Stay diversified. And focus on what’s real—not just what’s exciting. Whether your priority is growth, income, or a combination of both, our team is here to help you plan for your family’s financial future — no matter where you are in your career or retirement journey.