AI Bubble: Is It Bursting, and When Could It Burst?

Is there an AI bubble, and is it bursting? The evidence for and against, what happened in 2026, China's AI stock crash and the warning signs to watch.

AskZyro Team7 min read
AI Bubble: Is It Bursting, and When Could It Burst?

Updated 18 November 2026. This is a living guide: we refresh it as the story moves. Nothing here is financial advice.

Is there an AI bubble — and is it bursting? It's one of the biggest questions in markets and tech. Trillions of dollars are pouring into chips and data centres on the promise that AI will pay off, and in autumn 2026 the cracks became impossible to ignore: a brutal post-election sell-off, a $3.5 trillion warning from Gartner, the shock resignation of Nvidia's CEO and a Chinese AI stock crash.

Here's what the AI bubble actually is, the evidence on both sides, what has happened so far, the warning signs to watch, and what a burst could mean for you.

What is the AI bubble?

A financial bubble is when prices rise far above what the underlying businesses can realistically earn, fuelled by excitement and the fear of missing out, until confidence breaks and prices fall sharply. The AI bubble debate asks whether the valuations of AI companies — chipmakers, cloud providers, data-centre builders and AI labs — have run ahead of the profits AI will genuinely generate.

Crucially, a bubble is about money, not whether the technology works. As Axios put it, AI can be real and still be a bubble. The internet changed the world, yet the Nasdaq still fell by roughly 78% between its March 2000 peak and October 2002 when the dot-com bubble burst.

Is AI a bubble? The case for and against

Why many think it's a bubble

  • Spending far outstrips revenue. Gartner expects AI providers to spend nearly $1 trillion on infrastructure in 2026 and another $6 trillion from 2027 to 2030. To cover that, financing and a 12% return, data-centre owners would need $8.5 trillion of revenue by 2030 — but Gartner forecasts no more than $3.6 trillion. It warns more than $3.5 trillion of spending could face write-downs, and that over 80% of "neocloud" providers will stop operating independently by 2030.
  • Circular financing. Nvidia has invested in, lent to and guaranteed loans for companies that then buy its chips. Critics such as the CEPR's AI Bubble Monitor argue this inflates demand compared with an arm's-length market.
  • Rising debt and costlier money. More of the build-out is funded by borrowing just as interest rates have risen, which makes far-off profits worth less today.
  • Cheaper "good enough" models. If open or low-cost models keep closing the gap, the premium prices frontier labs need may be hard to sustain. (Vox)

Why others say it isn't (or isn't yet)

  • Real, fast-growing revenue. Unlike many dot-com start-ups, the leaders are earning serious money. Nvidia reported third-quarter revenue of $115.84 billion in November 2026, beating forecasts, and guided to $130 billion next quarter. (CNBC)
  • AI labs are scaling quickly. Anthropic's IPO filing showed third-quarter revenue of $17 billion, up from $1.4 billion a year earlier, and an annualised run rate of about $81 billion by 31 October. (AFP)
  • Demand still exceeds supply. Hyperscalers keep spending: Amazon expects about $220 billion of capital spending in 2026 and "meaningfully more" in 2027. (NBC News)
  • Falls so far are corrections, not a crash. Even after the autumn sell-off, the Nasdaq was still up for the year — far from a dot-com-scale collapse.

Is the AI bubble bursting? What's happened in 2026

DateWhat happened
June–October 2026AI stocks slide from summer highs; one generative AI ETF loses about 35% and the Philadelphia semiconductor index falls by more than a third (Vox).
14 OctoberGartner warns $3.5 trillion of AI infrastructure spending could be written down by 2030.
26 OctoberReuters reports China's STAR50 AI and chip index has fallen 40% in under four months.
30 OctoberIntel plunges about 25% on weak guidance as Big Tech earnings reignite fears about AI returns (CNBC).
5 NovemberAfter Democrats sweep the US midterms, fears of a crackdown on data centres send the Nasdaq down 5.5% into correction, the S&P 500 down 4.4% and the Dow down 1,833 points — the worst day since the April 2025 tariff shock. The "Magnificent Seven" lose over $1 trillion and Nvidia falls more than 9% (CNBC). Japan's Nikkei and South Korea's Kospi each close down about 6% the next day (BBC).
12 NovemberNvidia CEO Jensen Huang resigns after reports of extramarital affairs, including one with a woman suspected by US officials of spying for China. CFO Colette Kress becomes interim CEO. Shares fall 13% at the open before closing down 4.3% (The Guardian; Nvidia).
17 NovemberNvidia beats revenue forecasts and lifts guidance, briefly calming nerves — though its shares are still down about 17% for the year (Reuters).
18–19 NovemberAnthropic prices its IPO at an expected $70–$78 a share, valuing it at up to about $1.4 trillion — below the $2 trillion it had reportedly hoped for — ahead of Nasdaq trading as "ANT" on 19 November (CNBC).

So, has it burst? Not by the usual definition. A bear market is a 20% fall from a peak, and parts of the AI trade — chip stocks, neoclouds, some AI ETFs — are already there. But the wider market has not crashed, and leading companies are still growing. What has clearly changed is the mood: investors now demand proof of profits rather than rewarding every new spending plan. Retail traders, who used to buy every dip, were net sellers during the 5 November rout (Axios).

China's AI bubble burst

If you want to see what a deflating AI bubble looks like, look at China. Reuters reported on 26 October 2026 that the STAR50 index of AI and semiconductor leaders had fallen 40% in less than four months, wiping out about $480 billion. According to Goldman Sachs, AI stocks made up 42% of China's market value and 57% of trading, so few investors escaped.

  • Robot maker Unitree, 8,000 times oversubscribed at listing, had halved in value since August.
  • AI model maker MiniMax was trading about 50% below its IPO price, while Z.ai fell 63% in under a month after peaking in June (Financial Times).
  • Beijing stepped in with central bank swap operations and cheaper financing for share buybacks, but the rebound did not hold (Business Insider).

There was no single trigger: sky-high valuations, tax bills, the threat of Western bans on Chinese data-centre equipment and state-backed investors cooling speculation all played a part. One telling detail: official figures put rack occupancy in Chinese computing centres at 71.4%, but industry estimates suggest only 30–37% of intelligent-computing capacity is actually used (Al Majalla). Even so, as several analysts note, a stock crash doesn't mean demand for AI is collapsing — it means investors overpaid.

When will the AI bubble burst?

Honestly, nobody knows — and anyone who claims to is guessing. Bubbles are usually only confirmed in hindsight, and they can last far longer than sceptics expect. Even bearish analysts admit they can't time the top. Rather than a date, watch these warning signs:

  1. Revenue versus spending. Is AI revenue growing faster than capital spending? Gartner's $8.5 trillion vs $3.6 trillion gap is the number to watch.
  2. Hyperscaler capex plans. Cuts or pauses from Microsoft, Alphabet, Amazon or Meta would hit the whole supply chain.
  3. Financing stress. Harder borrowing, failed listings or troubled neoclouds signal the money is drying up.
  4. IPO reception. How Anthropic and other AI listings trade is a live test of investor appetite.
  5. Utilisation. Are data centres actually being used, or sitting idle like some Chinese computing centres?
  6. Politics and power. Regulation, energy prices and local opposition to data centres can slow the build-out.

What happens if the AI bubble bursts?

  • Markets: AI-heavy stocks and funds could fall sharply. Because a handful of tech giants dominate major indices, pensions and index funds would feel it too.
  • Companies: weaker players — especially debt-funded data-centre and neocloud firms — could fail or be absorbed, as Gartner expects.
  • Jobs and spending: companies might slow hiring and cut AI budgets, and the wider economy could slow.
  • The technology: it would almost certainly survive. After the dot-com crash, the infrastructure that was built stayed and powered the next wave. Cheaper chips and data-centre capacity could even make AI more affordable.

What should you do?

We can't give financial advice, but a few principles are widely shared. Check how much of your pension or investments sits in a few AI-linked companies, keep your portfolio diversified and invest for your own time horizon rather than the headlines. If you're unsure, speak to a regulated financial adviser.

For your work, the useful question isn't "will the bubble burst?" but "which AI tools actually save me time or money?" Tools that deliver real value tend to survive downturns. Explore hand-reviewed options in Discover, research a market with our market research tool, or ask AI Chat to explain any of the terms in this guide. And if you're tired of low-effort AI content, read our guide to AI slop.

The bottom line

The AI boom has many classic bubble ingredients — enormous spending, circular deals, rising debt and valuations built on future promises. Autumn 2026 showed how quickly sentiment can turn, and in China, a burst has effectively already happened. But the leading AI companies are earning real, rapidly growing revenue, and a correction is not the same as a collapse. The likeliest outcome may be the one history suggests: a painful reset for investors, while the technology itself keeps moving forward.

Frequently Asked Questions

Find answers to common questions about this topic.

What is the AI bubble?

The AI bubble is the idea that the share prices and valuations of AI companies — chipmakers, cloud providers, data-centre builders and AI labs — have risen far beyond the profits AI will realistically generate, and could fall sharply if confidence breaks.

Is the AI bubble bursting?

Parts of it are under strain. In autumn 2026, chip and AI stocks fell into bear-market territory, the Nasdaq dropped 5.5% in one day on 5 November, and Nvidia's CEO resigned. But the wider market has not crashed and leading AI firms are still growing revenue, so most analysts call it a correction rather than a burst — so far.

Has the AI bubble burst in China?

Many analysts say yes. China's STAR50 index of AI and chip leaders fell about 40% in under four months in 2026, wiping out roughly $480 billion, according to Reuters. Others argue it's a valuation reset rather than a collapse in AI demand.

When will the AI bubble burst?

Nobody can reliably predict it, and bubbles are usually only confirmed in hindsight. Signs to watch include AI revenue versus spending, cuts to Big Tech capital spending, financing stress at data-centre firms, weak AI IPOs and low data-centre utilisation.

Is AI a bubble like the dot-com bubble?

There are similarities — huge spending ahead of profits and soaring valuations — but today's leaders such as Nvidia earn far more real revenue than most dot-com companies did. Like the internet, AI is likely to survive even if investors lose money in a crash.

What happens if the AI bubble bursts?

AI-heavy shares and funds could fall sharply, affecting pensions and index funds; weaker data-centre and neocloud companies could fail; and businesses might cut AI budgets. The technology itself would very likely continue, possibly becoming cheaper.

How much is being spent on AI?

Gartner estimates AI providers will spend nearly $1 trillion on AI infrastructure in 2026 and another $6 trillion between 2027 and 2030, but forecasts no more than $3.6 trillion in related revenue by 2030.

Something else to read

Good finds are worth keeping.

Save tools to a personal shortlist, keep guides for later, and pick them up on any device. Your free account keeps it all together.