The market just reminded investors of something important, especially when excitement starts to outrun reality: Ask tough questions.
I’m sure you noticed, but stocks have been volatile this week, especially software stocks. The iShares Expanded Tech Software Sector Fund (IGV) – which holds a basket of software companies – fell about 13% from Monday’s open to Thursday’s close.
The approximate “reason” was the introduction of a new AI model from Anthropic. The model goes beyond simple “chatbot” features. It proved itself capable of automating tasks typically done with enterprise software.
Put another way, investors are starting to ask the right questions.
The question isn’t whether artificial intelligence (“AI”) will change the world. That part is largely settled for most of us. AI is already reshaping software development, logistics, drug discovery, customer service, and a lot more.
Instead, the market started asking the more dangerous question: Who is actually making money from this… And who could potentially be put out of business?
Rubber Meets the Road
That shift from story to fundamentals is the defining moment in every technology cycle. And right now, we are right in the middle of it.
The last time most of us saw a shift like this was during the dot-com boom of the late ’90s. Investors started asking some tough questions then, too.
Questions like…
Is the Internet actually boosting sales?
Does it really make sense for a pet food delivery startup to blow its entire budget on a Super Bowl ad?
We’re paying how much for shares of Cisco?!
Investors decided they didn’t like the answers to those questions. And stocks crashed.
The good news is that today’s AI boom doesn’t look like the dot-com bubble. I’ll explain exactly why with hard data soon.
The not-so-good news is that markets are still priced as if a lot will go right. Both things can be true at the same time. If you zoom out and look at the whole market, valuations tell a clear story.
Today, the S&P 500 trades around roughly 22 times forward earnings. That’s not bubble territory, but it is meaningfully above long-term averages. At the late-2025 peak, the market pushed even higher – closer to 23 times forward earnings – before recent volatility pulled multiples back slightly.
For context, right before COVID-19 in early 2020, the S&P 500 traded closer to 18 to 19 times forward earnings. That was already considered a relatively full valuation. Now, we’re operating roughly 15% to 20% richer on a valuation basis than that pre-pandemic baseline.
That doesn’t automatically mean markets crash. It does mean there’s a smaller margin for error. When valuations are elevated, earnings execution becomes everything. Hype alone can’t carry stocks forever.
But here’s where it gets important.
Even though today’s market is expensive relative to recent history, it still looks very different from the late 1990s technology bubble.
At the peak of the dot-com era, many tech companies traded at astronomical multiples. Some of the biggest names had no earnings at all. Many had no real path to profitability. Capital was cheap. Growth was taken for granted. Investor discipline on fundamentals was almost nonexistent.
Today, the largest AI beneficiaries are already massively profitable businesses. They generate real free cash flow. Tens of billions of it. They have fortress balance sheets with great credit ratings. And they have proven products customers already rely on.
That doesn’t mean their stocks can’t fall. We’ve seen that in real-time lately. But it does mean this cycle is built on a very different foundation than the dot-com era.
I don’t think we are really facing the “Is AI a bubble?” question that people think we are. I believe this is nothing more than the next technology adoption cycle.
Where Are We in The Cycle?
Every major tech wave tends to follow three phases. Understanding this can make all the difference when it comes to knowing when to allocate, what the risk is, and keeping cool under pressure.
The first is discovery. A new technology appears, and investors bid up almost anything associated with it. Narratives, not financials, dominate. I’d argue this phase started with the introduction of ChatGPT in November 2023. We’re past that now.
The second phase is infrastructure. This is where the real money starts concentrating in companies doing the real work. Think hardware, specialty manufacturing equipment, computing power, data infrastructure, and mission-critical software. In my view, this is where we’ve been for the past 24 months or so.
The biggest early winners in AI have largely been companies selling the picks and shovels. That’s why companies making real products forming the backbone of AI, like Taiwan Semiconductor (TSM) and Micron Technology (MU), were included in my Intelligent Options Advisor service. Fundamentals start to matter at this stage, but hype is still in the driver’s seat.
The third phase is when the rubber meets the road. The market separates companies that are merely “AI associated” from those making real money. Who has the competitive advantage, and how valuable is it? This is where fundamentals take their rightful place as the top priority. This is the stage we are entering.
What to Look For
I believe investors and Wall Street are going to focus on the two questions going forward.
-
Where are the profits? Companies are spending heavily on AI infrastructure and tools. Investors are going to demand proof that those investments either reduce costs or increase revenue. The whole “it’ll pay off eventually” mantra isn’t going to fly anymore.
-
Who has durable margins? As AI becomes mainstream, margins naturally compress. Companies with the deepest moats take the bulk of gains. The rest struggle. The market is realizing not every company is growing to the moon. And now it’s starting the messy process of figuring out which is which.
This is also why parts of the tech market, like software, have seen sharper pullbacks recently. Look at the stock charts of the AI infrastructure names like ASML (ASML), Micron, and Taiwan Semiconductor (TSM) that make real profits and trade at reasonable multiples despite the huge run-up in their stock prices. It’s hard to even see a sell-off. Some are at 52-week highs as this is being written.
Now, look at Palantir Technologies (PTLR) and Snowflake (SNOW). Very different story.
A big mistake investors make during hype cycles like this trying to time sentiment instead of focusing on fundamentals. It won’t work once the cycle reaches this stage. As painful as it is, this process is not bearish. It’s markets doing their job. The good news is that those focused on fundamentals will be rewarded long term. That wasn’t necessarily the case in 2024 and 2025, but it will be in 2026.
Final Thoughts
The bottom line is this: the AI boom doesn’t look like the dot-com bubble. But investors still need to tread carefully as the market separates the wheat from the chaff.
Real technological change is happening. Billions in profits are being made, and they are likely going much higher. But expectations are also high, and the tolerance for poor execution is gone.
Cash flows beat stories. Moats beat momentum. Fundamentals decide winners (and losers).
Regards,
Stephen Hester
Analyst, Wide Moat Research
|
