Technology

Is AI Heading for a Dot-Com Style Burst?

AI is probably over-funded in the short run, but it is unlikely to vanish the way many dot-com companies did. A correction is a real risk. The technology dying is not. Big tech is pouring hundreds of billions of dollars a year into AI, while the money coming back is a fraction…

An AI chip inside a bubble with a pin about to pop it, beside a burst dot-com bubble from 2000. The title asks whether AI is heading for a dot-com style burst.

AI is probably over-funded in the short run, but it is unlikely to vanish the way many dot-com companies did. A correction is a real risk. The technology dying is not.

Big tech is pouring hundreds of billions of dollars a year into AI, while the money coming back is a fraction of that. Many people are asking the same question: is this the dot-com bubble all over again?

The case for a bubble

Spending is far ahead of revenue. Amazon, Alphabet, Meta, Microsoft and SpaceX have guided to roughly $800 billion of capital spending in 2026, mostly for AI (HyperAI). By one estimate, AI services brought in only about $25 billion in 2025, around a tenth of what was spent on infrastructure (Invezz).

  • Cash is getting tight. Barclays expects Meta's free cash flow to fall nearly 90%, and Amazon is projected to go free-cash-flow negative this year (source).
  • Chips age fast. Companies write GPUs off over 5–6 years, but they may be outdated in 2–3. If so, profits today look better than they really are.
  • Circular deals. Some AI firms invest in each other and then buy from each other, which can make demand look bigger than it is.
  • Nervous markets. The Nasdaq had its worst day since April 2025 on 5 November, then jumped 3.4% four days later (source).

Why this isn't 2000 again

The biggest difference is who is paying. In 2000, the money was mostly debt and IPO cash going to startups with little or no revenue. Today, most AI spending comes from very profitable tech giants paying out of their own pockets (Nadcab).

Dot-com (1995–2000)AI boom (2023–2026)
Who funds itDebt and IPOs for young startupsCash from profitable giants
Real demandMany companies had little revenueAWS grew 36.7% in Q2 2026; Alphabet revenue up 25%
Rise in investment as share of US GDPAbout 1.2 pointsUnder 0.4 points so far

Sources: theCUBE Research, New Space Economy, Rework.

What a burst would look like

The likely outcome is a correction, not AI disappearing. Stock prices would fall hard. Weak AI startups and heavily indebted data-center operators would fail, and the giants would cut spending. One analysis puts the base case at a slow, volatile slowdown over the next year, with about a one-in-four chance of a full bust (India Macro Indicators).

The dot-com crash is a useful reminder. Pets.com died, but the internet didn't. The fiber-optic cables laid during the bubble became cheap infrastructure for the next 20 years, and Amazon and Google came out stronger. An AI crash would likely leave behind a lot of cheap computing power in the same way.

What it means for businesses building with AI

For companies that use AI rather than build models, a correction could help. AI costs would likely drop as spare capacity is sold off. The real risks are elsewhere:

  • Don't depend on one provider. Build so you can switch models if one raises prices or shuts down an API.
  • Sell results, not "AI-powered". Every AI feature should save customers time or bring them money they can measure.
  • Watch your margins. Know what each AI request costs you, and make sure your pricing covers it.

When bubbles burst, the companies that survive are the ones solving real problems profitably. That was true in 2000, and it will be true again.