Is This the AI Bubble, or Just the Market Taking Its Profit?
The market has a brutal way of asking questions.
It does not raise its hand. It does not clear its throat. It just sells.
That is what happened when South Korea’s KOSPI fell nearly 10%, triggering a trading halt and forcing investors to look again at one of the biggest stories in global finance: the artificial intelligence boom.
For months, AI has carried the kind of market energy that makes people nervous and excited at the same time. The technology is real. The spending is real. The demand for chips, data centres, cloud infrastructure and computing power is real.
But markets do not only price reality.
They price expectation.
That is where things get dangerous.
South Korea’s market has been lifted by the same global force driving US tech stocks: the belief that AI will need endless chips, endless infrastructure and endless computing power. Samsung Electronics and SK Hynix sit close to that story, which is why they helped pull the market upward.
Then they helped pull it down.
A 10% fall is not a whisper. It is a siren.
Francis Herd explains in Number of the Day that trading halts are designed to slow panic. When prices fall too fast, investors can stop thinking and start copying. One sell order becomes a crowd. A crowd becomes a stampede. A stampede can punish even good companies simply because fear is moving faster than analysis.
That is why this moment matters.
Not because one bad day proves AI is a bubble. It does not.
But because one bad day can expose how much pressure has built beneath the excitement.
Gareth Edwards asks the question directly: Was this a bubble bursting?
The comparison to the dot-com era is easy to understand. Back then, internet enthusiasm pushed valuations far ahead of the businesses underneath. Anything linked to the web could attract money. Then reality arrived, and companies without real earnings, real customers or real models were exposed.
AI has shadows of that story.
There is hype. There is crowding. There is fear of missing out. There are companies being priced not only for what they are, but for what investors hope they might become.
But this is not a perfect repeat.
Today’s AI economy is also built on major companies that already work. Nvidia sells the chips. Amazon and Google sell cloud infrastructure. Meta has a huge advertising machine. These are not imaginary businesses waiting for a miracle. They are profitable, powerful and embedded in daily life.
So the real question is sharper than “is AI a bubble?”
The real question is:
How much perfection can the market afford to price in?
Because when a stock or market rises too quickly, it does not need a disaster to fall. Sometimes it only needs doubt.
A regulator’s warning. A leveraged trade under pressure. A few big funds taking profit. A sense that everyone is standing on the same side of the boat.
Then the tilt begins.
For South Africans, the KOSPI may feel distant. But the AI economy is not. It touches phones, banks, streaming, search, logistics, cloud systems, cars, servers and the invisible digital plumbing behind modern business.
When global tech shakes, the tremor travels.
Maybe this is a correction. Maybe it is profit taking after an extraordinary rally. Maybe it is the market trimming the froth before the next leg higher.
Or maybe it is the first sign that AI stocks are about to face the one thing hype cannot dodge forever:
A demand for proof.
That is why 10% matters.
It is not the whole story.
It is the first hard question.
Chapters Timeline
1 (00:00) 10%, And Not a Good 10%
2 (00:17) Trading Halted in South Korea
3 (00:21) Why Markets Stop Trading
4 (00:59) Tech Stocks Sell Off Worldwide
5 (01:08) Nvidia, Samsung and AI Chips
6 (02:01) Why South Africans Should Care
7 (02:57) Was This the AI Bubble?
8 (03:08) The Dot-Com Comparison
9 (03:53) Samsung Down 12%
10 (04:09) Why This May Be Different
11 (04:39) Correction or Something Bigger?
12 (05:14) AI Spending Faces the Test
13 (05:25) The Final 10% Warning
Catch up on all Number of the Day episodes here: https://www.enca.com/number-day-podcast