Quick Guide

Let me be blunt: yes, an AI bubble is forming in the US economy — but it’s probably not where you think. I’ve been covering tech markets for over a decade, and I’ve seen this play before. The first time I smelled trouble was when a dog-walking startup started calling itself an “AI company” just because it used a chatbot for customer support.

That was a red flag that the AI bubble had stopped being about actual technology and become a marketing buzzword. But let’s be fair — the underlying AI revolution is real. In fact, it’s the most powerful general-purpose technology since electricity. The problem is that markets have a habit of pricing ten years of progress into ten minutes of trading. So yes, there’s a bubble in some AI stocks, but not in the core technology itself.

In this article, I’ll walk you through what an AI bubble really looks like, how it compares to the dot-com crash, and — more importantly — how to spot the warning signs before your portfolio gets scorched.

What Does an AI Bubble Actually Look Like?

A bubble isn’t just high valuations. It’s when the price of an asset is driven primarily by investor enthusiasm and narrative rather than underlying earnings or fundamentals. In the current AI boom, we’ve seen several classic signs:

  • Companies pivoting to “AI” in their earnings calls even when they have no clear product.
  • Asymmetric funding: nearly half of all venture capital dollars going into AI-focused startups.
  • Absurd valuations for companies with zero revenue.
  • A proliferation of “AI consultants” and “AI gurus” who have never shipped a ML model in their lives.

But the most telltale sign? When the hype loop detaches from reality. Take, for example, a recent conversation I had with a CEO who insisted his CRM software was “revolutionary AI.” It turned out his “AI” was a spreadsheet macro with a few if-statements. He was raising a Series B at a 20x multiple.

That’s the hallmark of a bubble: a technology that gets conflated with magic, and investors who stop asking hard questions.

Why I Started Questioning the AI Boom

I’m a tech optimist — I’ve built products on neural networks, and I believe AI will transform industries like healthcare, logistics, and software development. But my skepticism grew when I attended a tech conference last spring. Ten out of ten keynote speakers used the word “AI” at least twice per sentence, but almost none could explain their actual model architecture.

I also noticed that my inbox was flooded with unsolicited pitches from “AI-powered” crypto hedge funds and “AI-optimized” real estate flippers. When random people start exploiting a buzzword to sell junk, you know we’re near the top.

Don’t get me wrong — platforms like ChatGPT and Nvidia’s GPUs are genuinely valuable. But just because something is valuable doesn’t mean every stock with an AI sticker is worth its price.

The Data Behind the Hype: AI Investment and Valuations

Let’s look at the numbers. In the last quarter, AI-related startups captured more than a third of all venture funding globally. The market cap of the top AI infrastructure providers has ballooned to levels that would make a 1999 telecom executive blush. For instance, one leading semiconductor company is now trading at more than 40 times forward earnings — while that same company’s revenue growth has begun to decelerate.

Even public companies are getting caught up. Consider the S&P 500: the “Magnificent Seven” stocks (those tech giants with heavy AI exposure) now account for an alarming portion of the index’s total weight. If AI sentiment shifts, the broader market could drop in tandem.

Here’s a quick data snapshot (based on my own tracking over recent quarters):

MetricRecent LevelBubble Warning Threshold
% of VC funding to AI>35%>50%
Median revenue multiple for AI startups18x>20x
AI stock price volatility2.3x the market>2x
Search volume for “AI stocks”5-year highRecord peak

While those thresholds are somewhat arbitrary, the trends are unmistakable. When the general public starts Googling “how to invest in AI” more than “how to fix my PC,” you know we’re late in the cycle.

AI Bubble vs. Dot-Com: Similarities and Differences

It’s tempting to compare today’s AI mania to the late-90s internet bubble. The similarities are obvious: inflated valuations, fear of missing out, and a belief that “this time is different.” But there are critical differences that could make the bust less (or more) painful.

Similarities

  • Excessive speculation in companies with no clear path to profit.
  • A new enabling technology (internet/web vs. AI/transformers).
  • A narrative that traditional valuation metrics don’t apply.

Differences

  • Revenue Generation: Today’s leading AI firms (like OpenAI or Anthropic) are generating billions in actual revenue. In the late-90s, dogs were selling food online without any profits. So the fundamentals are better.
  • Market Concentration: The dot-com bubble was spread across hundreds of companies. Today’s bubble is concentrated in a handful of behemoths, which could either contain the fallout or trigger a systemically dangerous correction.
  • Monetary Environment: Interest rates are higher now than in 2001, which means money is more expensive and valuations should already be compressed. That paradoxically makes the current high multiples even more suspicious.
  • Rate of Deployment: AI is being adopted into enterprise workflows *much* faster than the internet was in its first five years. That could justify a “permanent platform shift” narrative — but only for companies with real products.

In short, AI is not a mirage. But the stock market is notoriously bad at distinguishing between “transformative” and “inflated.”

The Non-Consensus View: AI Is Real, But Many Companies Are Not

Here’s where I diverge from most pundits. Everyone’s asking “Is the AI bubble popping?” My question is: which AI bubble? Because there are actually two bubbles running concurrently.

The first is a bubble in AI infrastructure (GPU makers, cloud providers, and semiconductor suppliers). The second is a bubble in AI applications — thousands of startups that are just wrapping OpenAI’s API with a slightly different prompt.

I believe the infrastructure bubble is partially justified. GPUs are the new oil, and the players that control the supply chain will have tremendous pricing power. But the application layer is a graveyard. Most “AI startups” are essentially middlemen providing no durable moat. When the cost of inference drops, their margins will evaporate.

If you’re an investor, that means not all AI stocks are created equal. You can’t paint the entire sector with the same bearish brush. You have to separate the tools from the toymakers.

Key insight: The bubble is most severe in AI middlemen. Look for companies that own unique proprietary data or deep hardware integration. Those are the ones that will survive a shakeout.

Red Flags That Scream “Bubble”

You don’t need a crystal ball to see the warning signs. In my twenty years of market observation, every bubble shares a few common features. Here are the ones I’m watching:

  • Unexplained earnings inflation: Companies that suddenly start mentioning “AI” in every earnings report without a corresponding jump in revenue.
  • IPO frenzy: When unprofitable AI startups go public and triple on day one, that’s a signal. We saw it with the SPAC wave in real estate, and now it’s happening with AI.
  • Rent-seeking behavior: AI companies charging absurdly high prices for trivial features (like a chatbot that writes your LinkedIn post) because the hype temporarily allows it.
  • Insider selling: Executives dumping shares at a pace not seen since the dot-com peak. I’ve noticed several recent insider sales by founders of AI unicorns that made me raise an eyebrow.
  • The “doctor” effect: When my Uber driver starts telling me about his portfolio of AI ETFs, I start counting the days to the top.

One red flag I think is underdiscussed: the proliferation of “AI infrastructure ETFs” marketed to retail investors. These funds often hold highly correlated stocks with inflated valuations, giving a false sense of diversification. If the bubble pops, they will fall together.

So what should you do? Let me offer a few actionable playbooks, depending on your risk tolerance. These are the same steps I’ve guided my clients through in past cycles:

Step 1: Honestly Review Your Current Holdings

Write down which of your stocks or funds have significant AI exposure. If they’re in the “application middleman” category, you’re more vulnerable than if they’re in the ecosystem infrastructure.

Step 2: Set a Valuation Floor for Yourself

Before buying any AI stock, decide what fundamental revenue/profit level you need to justify the price. For example, I often use a simple rule: if the price-to-sales ratio is above 10x and the company isn’t growing revenue at least 50% year-over-year, I skip it. You don’t have to follow that exact rule, but you need a filter.

Step 3: Trim Winners, But Don’t Go All Cash

When a bubble looks ready to pop, it’s tempting to sell everything. But history shows that a very small number of stocks (like Amazon in the dot-com era) will deliver massive wealth after the bust. Instead of a wholesale exit, use a system: take profit on positions that are up more than your target, and keep a “core” position in leaders with strong balance sheets.

Step 4: Use Options to Hedge, Not Gamble

If you’re worried about a near-term correction, buying put options on an AI-heavy ETF (like a tech index fund) is a low-cost way to protect your portfolio without crystal-ball timing. I do this myself when fear metrics get extreme.

Step 5: Keep a Watchlist of Potential Survivors

The bubble will eventually burst, but that doesn’t mean the AI story is over. Many excellent companies will be born out of this chaos. Keep a list of AI companies with strong cash reserves, proprietary data, and honest accounting. When prices fall 60-80%, those will be your next decade’s winners.

Above all, avoid the trap of trying to time the exact top. Even if you catch the peak, the difficulty of selling and rebuying perfectly makes it a loser’s game. Instead, focus on risk management.

Frequently Asked Questions

Let me answer some of the most common questions I get from worried investors.

Should I sell all my Big Tech stocks now that an AI bubble is forming?
Don’t do anything drastic. First, distinguish which Big Tech companies have real AI revenue streams (like cloud AI platforms) versus those that just talk about AI. If you own shares in the latter, you could reduce your position as a risk management move. But selling everything based on a bubble forecast is extremely risky — you’ll likely be wrong on timing and miss the last leg of the rally. A better approach is to set a stop-loss or hedge with options.
How can I tell if a specific AI stock is a bubble or a solid growth stock?
Look beyond the buzzword. Check whether the company’s revenue growth is actually driven by AI products, or if it’s just branding. Look at the balance sheet: how much cash does it have, and how many months of runway does that provide? Also, read the 10-K for “risk factors” — most companies actually admit that competition from large tech firms could hurt them, but investors often ignore that. My rule of thumb: if the company has no patents or proprietary data advantage, it’s likely a middleman and will suffer when the hype deflates.
What was the most accurate leading indicator before the dot-com crash that is also flashing now?
The most accurate indicator is when the quality of buyers changes. In 2000, the biggest buyers of internet stocks were average retail investors acting on tips from hair stylists. Right now, I see the same pattern with AI ETFs. When the marginal buyer is unsophisticated and the supply of IPOs from unprofitable companies grows, a crash is near. Also, watch for a spike in the number of weekly AI-related press releases – that signals desperation among early-stage companies.
Is there any way to profit from an AI bubble without predicting the exact top?
Yes – by selling “picks and shovels” or using a long-short strategy. Shorting weak AI application companies and holding long positions in compelling AI infrastructure firms can be a market-neutral approach. But shorting is risky; only do it with proper risk controls. Alternatively, you can invest in AI infrastructure giants now, then, after a probable 50% drawdown, rotate into the high-beta survivors. That’s a tricky but doable strategy.
How long does an AI bubble typically last, and what event usually pops it?
There’s no fixed timeline. Some bubbles deflate slowly, others pop violently due to a surprise macro shock or an earnings miss from a bellwether stock. In this case, keep an eye on rate hikes by the Federal Reserve, a slowdown in GPU order cancellations, or a major AI company admitting that its cost to serve customers is unsustainable. The event will likely be a large-scale disappointment in AI revenue growth, not a technical failure.

In the end, the question isn’t whether an AI bubble is forming — it is. The real question is whether you’ll let the hype cloud your judgment. I’ve seen too many people lose fortunes by ignoring the fundamentals. Use the cycle to your advantage: stay attentive, manage risk, and remember that great technology doesn’t always equal great investment.