AI Bubble or Not? What the 2000 Dot-Com Crash Gets Wrong (and What Investors Should Watch Instead)

Every week we sit down with you to discuss what’s moving the financial markets, based on our investment approach.  Sometimes we focus on individual stocks, sometimes on sectors, and sometimes on the bigger macro picture.

This week, we couldn’t avoid the question that keeps popping up everywhere:

Is AI a bubble… or not?

It’s a fair debate, especially when markets feel jittery and investors start searching for “signals” that things are overheating. And yes, the comparison comes up every time: “This looks like the internet bubble of 2000.”

But here’s the key point: AI is not “the bubble” in the same way the internet wasn’t “the bubble.” The bubble (if there is one) happens in parts of the market—in valuation, financing structures, and unrealistic expectations; not in the underlying technology itself, which is why long-term investing principles matter more than headlines.

Let’s break it down in a way that’s actually useful from an investor perspective.

AI Isn’t Going Away (So That Alone Isn’t a Bubble)

A bubble is usually framed as “something disappears and leaves nothing behind.”

AI doesn’t fit that. The technology is already being deployed across industries, and adoption is increasing. In that sense it’s similar to the internet: the internet wasn’t a bubble—the pricing and speculation around certain internet companies was.

So the better question is:

Where is the money flowing, and what assumptions are investors paying for?

Our Framework for Thematic Investing: 3 Ways Investors Can Get Exposure to AI

This structure reflects our framework for thematic investing, focusing on where value compounds over time rather than chasing hype. To keep this practical, we split the “AI investment universe” into three categories:

1) The “Picks and Shovels” Companies (The Infrastructure Layer)

Think of the old gold rush: the people who often did best weren’t the ones digging for gold, they were the ones selling the tools.

In AI, this includes businesses that enable AI at scale: chips (CPU/GPU and related hardware), data centres, networking, energy and power infrastructure, connectivity and telecom capacity (a modern parallel to early internet adoption).

This layer has benefited massively because AI computing demand is expected to keep rising, while also supporting diversifying across sectors. The real debate is not if it grows, but how far and how fast.

2) Companies Already Using AI to Improve Profitability (The Adoption Layer)

This is where it gets interesting, because these companies can potentially benefit from AI without needing “AI hype” to work out.

These businesses use AI to: reduce costs, improve productivity, increase revenue per employee, and automate processes and decision-making.

In the captions, one example mentioned is how large firms use AI-driven systems to detect issues automatically (think: instant inspection, automated quality checks, and operational efficiency). The point is simple:

If Nvidia falls 3% tomorrow, these companies can still be improving margins quietly in the background.

3) The Companies Building AI Applications (The Product Layer)

This includes the firms creating AI tools, platforms, and end-user applications (healthcare is one example mentioned).

This layer can be powerful, but it’s also where expectations can get ahead of reality, because the addressable market is huge, and that tempts investors to price in “perfect execution.”

Nvidia as the Barometer: Great Results… and Still Volatility

Nvidia remains a market barometer for AI sentiment. In your captions, the market reaction is a perfect example of how fragile positioning can be: strong numbers (revenue and profit beating expectations), an initial rally, then the move reversed and the stock closed down.

That pattern matters. Not because it “proves” AI is a bubble, but because it shows the market is constantly repricing expectations.

The core valuation issue isn’t “Nvidia is good or bad”

It’s this:

What growth is the market already paying for, and does that growth need to be flawless for years?

When investors price in big earnings expansion over a multi-year period, the stock becomes sensitive to anything that hints at slower growth, changing payment cycles, rising inventories, or tougher competition.

The “Circular Money” Risk Investors Should Understand

One risk raised in the discussion is the idea of a feedback loop in the AI ecosystem: AI labs and growth companies raise or borrow money, they spend heavily on compute and chips, the infrastructure suppliers report booming demand, which fuels more investor enthusiasm and more funding…

That can be fine in a genuine growth cycle. But it becomes risky if, funding dries up, payment cycles stretch, or the end-demand doesn’t mature as fast as expected.

This is where “bubble-like” dynamics can appear—not in AI adoption itself, but in how aggressively the ecosystem is being financed.

The Dot-Com Comparison: The Numbers Are Very Different

The internet crash is the favourite comparison, so let’s address it directly.

Back then (1999–2002)

The problem wasn’t that the technology was fake. The problem was: many companies had little or no profit, often heavy losses, valuations relied on distant, uncertain future earnings, and the market’s pricing assumptions were extreme.

Today

The major difference highlighted in your discussion is crucial:

Today’s mega-cap tech businesses are generating real, record-level profits and strong margins.

That doesn’t mean prices can’t correct. It does mean the market structure is different from an era where huge parts of the “new economy” didn’t make money at all.

So when people say, “This must crash like 2000,” they often overlook lessons from past market crashes and they’re skipping the most important part: earnings quality.

So… Is AI a Bubble?

If we define a bubble as “it explodes and nothing is left,” then AI itself is not a bubble.

A more realistic conclusion is that AI is a structural growth theme (the technology stays), some valuations can still be too optimistic (stocks can fall hard), and the most fragile players will be the ones with high debt and weak cash flows if growth expectations disappoint or financing tightens.

In other words: the risk is not “AI disappears.” The risk is who survives a slowdown, and which expectations were unrealistic.

Investor Takeaways: What to Watch Next

If you’re investing around the AI theme, here are the practical checks that matter: Earnings and margins: are profits real and durable?

Cash flow vs. reported profit: do the numbers translate into cash?

Balance sheet strength: low debt + strong cash gives resilience.

Customer concentration and payment cycles: are buyers taking longer to pay?

Competitive pressure: dominance rarely stays permanent in fast-growth tech.

Diversification across the 3 AI layers: don’t bet everything on one “barometer” stock.

Final Thought

The internet changed the world and investors still lost money buying the wrong companies at the wrong prices.

AI can be the same story: real technology, real winners, real losers, and very real volatility in between.

If you want help positioning your portfolio with sensible exposure (without making a single-stock gamble), get in touch with us, or explore our other articles and videos on technology investing, sector ETFs, and portfolio diversification.