Are We in an AI Bubble? An Investor’s Honest Take

The S&P 500 tech sector is cheaper today than it was in January, and almost nobody is talking about it. That single fact reframes the entire bubble debate. 

Stocks won’t stop going up. Even an oil spike and a Middle East war couldn’t slow them down. Now everyone wants to know if the AI trade has finally gone too far. 

Writing this the week of Micron’s blowout quarter, my answer is no. We are not in an AI bubble. We are in an earnings boom that happens to be powered by AI, and there is a meaningful difference between the two. 

After three phenomenal years for the market in 2023, 2024, and 2025 coming out of the October 2022 lows and the launch of ChatGPT, investors large and small are asking the same question. All the market seems to do is go up and to the right, and everyone is saying stocks are crazy expensive. 

There is something to that. On the surface, stocks aren’t cheap. But they aren’t crazy expensive either. 

Is This 1999 All Over Again? 

No, and the comparison breaks down quickly when you look under the hood. 

In 1999, the leaders had no earnings, negative cash flow, and balance sheets propped up by venture money. Today’s AI leaders are some of the most profitable companies in human history. Nvidia, Microsoft, Google, Meta, and Apple generate hundreds of billions in free cash flow combined. Most of them pay a dividend. Their balance sheets are fortresses. 

In 1999, the Nasdaq traded at more than 70x forward earnings at the peak. Today the S&P 500 Information Technology forward P/E (the price you pay today for one dollar of next year’s earnings) sits around 21.8x, actually lower than the 25.3x it started the year at. Nvidia trades around 20x forward earnings, well below its five year average near 36x. These are not bubble multiples. These are reasonable multiples on companies growing earnings 20 to 30%. 

The dot-com bust was a story of valuations detached from fundamentals. Today’s story is fundamentals catching up to and in some cases outrunning valuations. 

The S&P 500 grew earnings roughly 14% in 2025, well above the long term annual average near 9%. Consensus is calling for about 18% earnings growth in 2026, with Q2 2026 already tracking near 26% year over year. 

This is the earnings boom, not the AI bubble. 

Bubbles are built on hope and hype. This market is being built on profits, cash flow, and capex that is actually showing up in the income statements. 

The rally has broadened from GPU bottlenecks at Nvidia, to the hyperscalers (the mega-cap cloud providers like Microsoft, Amazon, Google, and Meta who buy AI chips in bulk), to memory stocks like Micron, with AI adjacent areas of the market booming as well. 

Micron is the cleanest example of why this isn’t a bubble. 

A forward P/E of 8x on a company doubling revenue. Apply a normal market multiple to those earnings and the stock is nearly 3x higher than it trades today. That is not a bubble. That is a market still trying to figure out how to price a structural shift in memory demand driven by high bandwidth memory (HBM, the specialized chips inside every AI server). 

What About the Bear Case? 

Healthy skepticism is warranted. Here are the loudest concerns making the rounds on podcasts, in newsletters, and across financial Twitter, and where I land on each. 

The deals between Nvidia, OpenAI, AMD, Oracle, and CoreWeave have created a web where the same dollars appear to be flowing in a loop. It’s worth watching, and it is one of the single most legitimate red flags in the market right now. But these are real chips going to real data centers running real workloads with real revenue on the other side. That is not Pets.com selling kibble at a loss. 

There was an MIT study that showed that 95% of enterprise AI pilots fail. True, and exactly what you would expect this early in an adoption curve. The same was true of early internet projects, early cloud migrations, and early mobile apps. The 5% that work are already generating enough productivity gains to justify the entire capex cycle. 

The market is extremely concentrated. Concentration risk is real risk. Index investors own more tech than they realize.  

This is the better question, and almost nobody is asking it. Are we in an earnings bubble, not an AI bubble? 

The long term annual earnings growth rate for the S&P 500 has averaged about 8 to 9% over the last decade. We have been running well above that pace for three years, with about 11% in 2023, 8% in 2024, and 14% in 2025, and consensus calling for roughly 18% in 2026. 

Is that sustainable forever? Probably not. Earnings growth eventually mean reverts. But with AI infrastructure capex projected to climb from roughly $750 billion this year to more than $900 billion next year across the broader ecosystem, the earnings engine has real fuel behind it for at least the next 12 to 24 months. 

The right framing for this market isn’t “AI bubble.” It’s “earnings boom with an expiration date we can’t yet see.” 

Emerging market stocks are flying, buoyed by chip manufacturers in Korea. The KOSPI is up nearly 98% year to date, with Samsung Electronics and SK Hynix leading the charge thanks to their dominance in high bandwidth memory used in AI servers. 

When the rally broadens to Seoul, Taipei, and beyond, it stops looking like a narrow speculative mania and starts looking like a global capex cycle. Bubbles narrow. Booms broaden. 

There are a few things would move me from “ride it” to “raise cash.” 

Hyperscaler capex guidance flattening or rolling over. Forward earnings estimates getting revised down two quarters in a row. Forward P/Es on the leaders pushing into the mid 30s without matching earnings growth. Credit spreads blowing out while equities keep climbing. Real signs that AI revenue at the customer level isn’t showing up. 

None of those are flashing red right now. A couple are amber. Most are green. 

So, Are We in an AI Bubble? 

I tend to think not. Bubbles are built on hope and hype. This market is being built on profits, cash flow, and capex that is actually showing up in the numbers. Could it get stretched from here? Of course. Will there be a correction at some point? Always. But there is a big difference between a market that needs to cool off and a market that is about to come unglued. Right now, the earnings say this one still has legs. 

And if I’m wrong and this turns out to be a bubble, it’s one I plan on riding for a while longer. 

What would have to happen for you to call this a bubble? 

Austin Wilson is a Partner and Chief Investment Officer at Hixon Zuercher Capital Management, a registered investment advisor in Findlay, Ohio managing over $500 million for over 370 families and institutions.

Disclaimer: The views expressed in this blog are solely those of Austin Wilson and do not necessarily reflect the views of Hixon Zuercher Capital Management. Hixon Zuercher Capital Management and its clients may hold positions in securities mentioned. Nothing in this blog constitutes investment advice or a recommendation to buy or sell any security. All investments involve risk, including possible loss of principal. Please consult your financial advisor to ensure any investment aligns with your objectives and risk tolerance. Indices are unmanaged and not directly investable. Any investment in securities, funds, or other financial instruments will involve costs and fees, including but not limited to fund expense ratios, management fees, and transaction costs. This content is provided for informational purposes only and is not intended to comply with the requirements of the Investment Advisers Act of 1940 or any other applicable securities laws. Past performance is not indicative of future results.