AI Bubble: Is There One? A Plain-English Explainer (2026)

You have probably heard the question a hundred times by now: is AI a bubble? With trillions of dollars flowing into artificial intelligence and stock prices soaring, plenty of people are asking whether the whole thing is about to burst. Here is the short answer.

The quick answer: There are real signs of a bubble in 2026 — huge spending running ahead of revenue, sky-high valuations, and sharp sell-offs in AI stocks. But today’s AI leaders also earn massive, genuine profits, which makes this very different from the dot-com crash of 2000. Honest experts are split, and no one can say for sure whether it pops or when.

A quick note before we start: this article is general information, not investment advice. Do not make financial decisions based on it — talk to a qualified adviser if you need advice about investing.

What people actually mean by “AI bubble”

A bubble happens when the price of something climbs far above what it is really worth, because everyone expects the price to keep climbing. At some point the money stops flowing, prices collapse, and people who bought at the top lose a lot.

When people say “AI bubble,” they usually mean one of four different things, and mixing them up causes a lot of confusion:

• The technology itself. Is AI real and useful, or overhyped? Almost everyone agrees the technology genuinely works — it is being used in businesses everywhere.
• Enterprise results. Companies are spending big money on AI tools. Are they getting enough value back for it?
• Infrastructure payback. Tech giants are spending enormous sums building data centres and buying chips. Will that spending pay for itself?
• Stock prices. Are AI company shares priced fairly, or has excitement pushed them too high?

Analysts at IoT Analytics made exactly this point in a September 2026 report, arguing these are four separate questions and the infrastructure build-out is the most important one.

The case that it IS a bubble

The bubble camp has real ammunition. Here is their strongest evidence:

The spending is enormous

The big tech companies are pouring money into AI infrastructure at a speed the world has never seen. Morgan Stanley estimates that five hyperscalers — Alphabet, Amazon, Meta, Microsoft and Oracle — will spend a combined $805 billion on capital expenditure in 2026 alone, according to reporting by the South China Morning Post.

The Bank of England has noted that estimates for hyperscaler AI capital spending in 2028 jumped from under $600 billion to more than $1 trillion in less than a year. That is a projection nearly doubling in months, not years.

Revenue has not caught up

The central worry is simple: the money being spent today is running far ahead of the revenue it has produced so far. Alibaba chairman Joe Tsai warned last year that he was “starting to see the beginning of some kind of bubble,” pointing to data centres being built on speculation without secured customers.

Macquarie’s head of strategy Viktor Shvets told a conference in 2026 that an AI bubble “absolutely exists at the low-end infrastructure end” — the chips, data centres and hardware layer — where investors are piling money into a transformative technology long before revenue catches up.

The warnings are coming from serious places

The International Monetary Fund’s October 2026 Global Financial Stability Report names an AI market correction as one of the biggest risks to the entire global financial system. The European Central Bank compared the current AI rally with the late-1990s technology boom in August 2026, questioning whether today’s valuations reflect rational expectations or another speculative cycle.

And there have already been scares. In July 2026, AI company shares “fell sharply,” and the Bank of England’s Financial Policy Committee record says the fall was made worse by leveraged investors being forced to sell. On October 14 and 15, AI stocks fell again after US inflation came in hotter than expected — CoreWeave dropped 15% in a single day and Oracle fell nearly 9%.

Some companies already look bubbly

Research from Cornell University, published in August 2026, found that the AI sector as a whole is not a bubble — but some individual AI companies, including Google’s parent Alphabet, are showing price behaviour consistent with a bubble. As Professor Martin Wells put it, treating “AI” as one single category hides big differences between firms.

The case that it is NOT a bubble

The other side is not just wishful thinking. They have their own solid points:

The profits are real

This is the biggest difference from the dot-com era. In 2000, companies with no revenue and no profit were worth billions. Today, the companies leading the AI boom are wildly profitable. Nvidia reported quarterly revenue of $96.2 billion for the quarter ended July 2026 — up 106% from a year earlier — with its data centre business alone generating $89 billion. The big cloud companies are funding their AI spending from actual cash, not debt.

Money keeps flowing in

A bubble pops when buyers run out. That has not happened. The sell-offs in 2026 — in July and again in October — were followed by recoveries. The money pouring into AI has not stopped.

Bubbles in real technologies often take years

Macquarie’s Viktor Shvets argues the AI infrastructure boom is unlikely to burst in 2026 or 2027, comparing it to past bubbles in general-purpose technologies like the internet and telecoms. His view: we may still be early in the cycle, because there is no sign yet of the things that actually end booms — unsustainable debt, cash running dry, or valuations completely detached from reality.

AI capability is still improving

Analyst Benedict Evans summed up an important point: “The fact that AI is working really, really well, and has no signs of hitting a wall, and will change the world… does NOT mean that there cannot also be a bubble in AI.” Real technology and a real bubble can exist at the same time.

AI bubble vs. the dot-com crash

The comparison everyone makes is the dot-com bubble of 2000. Here is how they are similar, and how they differ:

• Similar: Huge excitement, massive infrastructure spending, valuations based on optimistic future growth, and sharp sell-offs along the way.
• Similar: Telecoms companies in the late 1990s laid far more fibre-optic cable than the market needed — a warning for today’s data centre building spree.
• Different: Today’s leaders earn huge real profits. In 2000, many of the hottest companies had never made a cent.
• Different: AI spending is funded mostly by cash-rich companies, not by borrowed money the way much of the dot-com boom was.
• Different: The internet genuinely changed everything despite the crash — the same may well be true of AI even if parts of today’s market are overpriced.

What to watch

If you want to judge for yourself where this is heading, here are the signals experts are watching:

• Whether the spending pays off. The key question is whether hundreds of billions in data centre spending produces matching revenue. IoT Analytics projects hyperscaler AI returns could drop below zero by 2031 — something the giants could absorb, but worth watching.
• Interest rates. Higher rates hit AI spending hardest because its payoff is years away. The US Federal Reserve raised rates to 3.75–4.00% in September 2026, and the 10-year Treasury yield hit 5.106% on September 24, its highest since 2007. Both sell-offs in 2026 were tied to rate worries.
• The most indebted corners. The cracks so far have been in the most leveraged parts of the boom — heavy borrowers building data centres — not in demand for AI chips, which remains strong.
• Concentration. Market gains are driven by a handful of names. When everything depends on a few companies, any stumble hits the whole market.

FAQs

Will the AI bubble burst in 2026 or 2027?

No one knows. Some analysts, like Macquarie’s Viktor Shvets, think it is unlikely to burst in 2026 or 2027. Others say sharp corrections are already happening. What everyone agrees on is that timing a pop is impossible.

Is Nvidia in a bubble?

Nvidia is the most debated stock in this conversation. Its revenue is enormous and still growing fast, which supports its price. But its valuation assumes that growth continues for years — and if AI spending slows, the stock is exposed. Cornell’s research suggests some individual AI stocks show bubble-like behaviour even if the sector as a whole does not.

Did the AI bubble already pop?

No. A bubble pops when prices fall and keep falling because the buyers have run out. In 2026, AI stocks fell hard twice and recovered, while spending kept rising. The boom is wobbling, not over.

What happens if it does pop?

Look at history: the internet survived the dot-com crash and changed everything anyway. If an AI bubble pops, the technology keeps working — but investors who bought at the top, and the most indebted companies, would suffer most. The IMF considers an AI correction a serious risk precisely because so much money is now tied up in it.

The bottom line

So, is there an AI bubble? The honest answer is: partly. Parts of the AI market — particularly infrastructure spending and some individual stocks — are showing genuine bubble warning signs, and serious institutions like the IMF and the European Central Bank are watching closely. At the same time, the technology is real, the profits are real, and the differences from 2000 are real too.

The smartest way to think about it is not “bubble or no bubble” but “where exactly is the froth, and where is the substance?” The data centres, the chips, and the stock prices are different questions — and they may have different answers.