Artificial Intelligence has become the talk of the town. From boardrooms to coffee shops, dinner tables to trading floors, AI dominates conversations and captures imaginations. And rightly so: we are living through what many call the Fourth Industrial Revolution.
The first three revolutions fundamentally reshaped human progress:
- Mechanisation powered by steam (~1800)
- Mass production enabled by electricity (~1900)
- Digitalisation culminating in the internet (~2000)
Today’s AI revolution stands firmly as the next seismic shift. Unsurprisingly, it is driving enormous excitement, and enormous investment. Companies that fail to ride this wave fear being swept aside by it.
It has now been three years since the “ChatGPT moment” of November 2022, widely viewed as the spark that ignited the current AI boom. And as with every major technological transformation, an inevitable question hangs in the air: Are we in a bubble?
The Nature of Bubbles, and Why Investors Secretly Like Them
Capitalism naturally drives capital toward significant waves of innovation. When the perceived opportunity is huge, money rushes in, often excessively, into both real businesses and the stock market.
So when people ask whether we are in a bubble, what they are really asking is: Is the bubble about to burst?
Because the truth is: Investors love bubbles. They just fear the endings.
Stock prices rise quickly during bubbles, often at breathtaking speed. What worries investors is the violent decline that follows.
And of course, bubbles contain layers of irrationality. There is no visible clock in the ballroom, so leaving the “Cinderella party” just before midnight is practically impossible. Emotionally, it’s also difficult to exit early, because the final stretch of a bubble is usually the most exhilarating, and the most profitable. Prices escalate sharply; this is the part of the night when the “lucky draw prizes” are handed out.
Still, people ask: How close are we to midnight?
I will offer my guess, though I could be completely wrong. Fortunately, our investment framework does not depend on predicting the exact timing of the bubble’s end.
My view: Yes, there is an AI bubble, but we may still be some distance from its bursting.
Looking Back: Lessons From the Dot-Com Bubble
A useful comparison is the Dot-com crash of March 2000.
1. Market Performance
Five years before the internet bubble burst, the Nasdaq rose ~550%. In the past five years, the Nasdaq is up ~150%. Large, but nowhere near the frenzy of 1995–2000.
2. Profitability
The internet pioneers of 1999 were largely unprofitable. By contrast, today’s AI leaders are hugely profitable.
The more speculative, unprofitable AI ventures are mostly still private, including OpenAI itself, which is expected to go public in 2H 2025. Nevertheless, if these private AI companies run into troubles, there will be contagion effect to the public ones.
3. Barriers to Entry
Starting an internet company in the 1990s required little capital. Starting an AI company today requires enormous resources, specialist talent, compute, data, resulting in much higher barriers to entry.
4. Smart Money Behaviour
Warren Buffett famously avoided internet companies during the Dot-com boom and refused to invest even at the peak of the excitement in 1999.
Recently, however, he invested US$4.5 billion in Alphabet, one of the central players in the AI revolution.
Taken together, today’s environment looks far less exuberant than the internet bubble at its peak.
The Greenspan Reminder
Alan Greenspan warned about “irrational exuberance” on December 5, 1996. Yet the bubble took another 3 years and 3 months to burst. During that time, the Nasdaq climbed another ~300%.
Being early is just as dangerous as being wrong.
Our Strategy: Participate, But Stay Grounded
Rather than trying to predict the exact bursting point, we rely on our timeless investment framework:
1. Keep AI exposure meaningful yet measured
Maintain a 5–15% allocation to AI-related equities, large enough to benefit, controlled enough to limit downside.
2. Rebalance regularly
As AI stocks rise, we take profits and channel them into:
- gold
- bonds
- high-quality companies in other sectors with good value
This ensures we stay invested without becoming overexposed.
Long-Term Success Doesn’t Require Crystal Balls
With this approach, we can confidently participate in the AI revolution—whether it continues for several years or begins to cool sooner than expected.
In the end, long-term success in investing doesn’t come from predicting the world without. It comes from cultivating the framework we build within.
