Is the AI Bubble over?

Not yet.

How we decide ↓

The AI bubble tracker

Is the AI bubble about to burst, or has it already? Instead of opinions, this page checks the hard signals a burst would leave behind: an Nvidia stock crash, a bear market in AI and tech stocks, Big Tech cutting its AI spending, and a glut of cheap GPUs. The data updates automatically from public sources, and the answer above changes the moment all of them are true.

The four criteria

The answer flips to Yes only when all four are met at the same time. Every number is pulled automatically from public data.

0 of 4 criteria met · updated

  1. 1. Nvidia stock crashes

    Not met

    NVDA trades at least 50% below its all-time high.

    Now: 0.0% below peak

    Nvidia is the purest public bet on AI. Every past tech bubble ended with its leader losing more than half its value (Cisco fell ~80% after 2000).

    NVDA at $242.11 vs. all-time high of $242.11 (2026-10-06). Source: Yahoo Finance

  2. 2. AI and tech stocks enter a bear market

    Not met

    The Nasdaq-100 is at least 30% below its all-time high.

    Now: 0.0% below peak

    A bubble bursting drags the whole sector, not just one stock. The Nasdaq-100 fell 33% in 2022 and 83% after the dot-com peak.

    Nasdaq-100 at 31,355 vs. all-time high of 31,355 (2026-10-06). Source: Yahoo Finance

  3. 3. Big Tech cuts AI spending

    Not met

    Combined quarterly capex of Microsoft, Alphabet, Amazon and Meta falls year-over-year.

    Now: +87% YoY

    Hyperscaler data-center spending is the money that funds the whole AI supply chain. When it shrinks, the boom has stopped.

    2026 Q2: $165.1B combined vs. $88.2B a year earlier. Source: SEC EDGAR filings

  4. 4. GPU glut: H100 prices collapse

    Not met

    Median on-demand H100 rental price falls below $1.00 per GPU-hour.

    Now: $2.47 / GPU-hour

    If GPUs become nearly free to rent, the world built more AI capacity than anyone is willing to pay for — the classic overbuild of a bubble.

    Median of 9 live H100 offers (range $1.25–$3.03). Source: Vast.ai GPU marketplace

AI bubble FAQ

Is the AI bubble over?

Not yet. As of October 6, 2026 at 03:15 PM UTC, 0 of 4 criteria that would signal the end of the AI bubble are met. The answer only changes to yes when all 4 are met at once.

Has the AI bubble burst?

No. A burst would show up as a crash in AI stocks, a deep tech bear market, falling Big Tech AI spending and a glut of cheap GPUs. As of October 6, 2026 at 03:15 PM UTC, 0 of these 4 signals are present.

Is Nvidia in a bubble?

Nvidia is the clearest public bet on AI, so its stock is our first signal. Right now NVDA is 0.0% below peak. NVDA at $242.11 vs. all-time high of $242.11 (2026-10-06). We would count it as a burst only at 50% or more below its all-time high; in 2000, Cisco, the leader of the dot-com boom, fell about 80%.

Are AI stocks crashing?

Not by our definition unless the tech-heavy Nasdaq-100 is at least 30% below its record. It is currently 0.0% below peak. Nasdaq-100 at 31,355 vs. all-time high of 31,355 (2026-10-06).

Is Big Tech still spending on AI?

Combined quarterly capital spending of Microsoft, Alphabet, Amazon and Meta is running at +87% YoY. 2026 Q2: $165.1B combined vs. $88.2B a year earlier. Spending that funds data centers and chips is the fuel of the AI boom; a year-over-year decline would be a major warning sign.

Are GPUs getting cheap? (H100 rental prices)

The median on-demand rental price of an Nvidia H100 GPU is $2.47 / GPU-hour. Median of 9 live H100 offers (range $1.25–$3.03). A price below $1 per GPU-hour would suggest the world built far more AI capacity than it can use.

AI bubble vs. dot-com bubble: how do they compare?

Both share record capital spending on infrastructure ahead of profits and a few market leaders carrying the index. The dot-com bust ended with the Nasdaq-100 down about 83% and leaders like Cisco down about 80%. Unlike 2000, today's AI leaders are highly profitable, which is why our criteria require the money flow itself (capex, GPU prices) to reverse, not just stock prices.

How do you decide whether the AI bubble has burst?

We track 4 measurable signals and update them automatically from public data: NVDA trades at least 50% below its all-time high; The Nasdaq-100 is at least 30% below its all-time high; Combined quarterly capex of Microsoft, Alphabet, Amazon and Meta falls year-over-year; Median on-demand H100 rental price falls below $1.00 per GPU-hour. Each one alone can happen in a normal correction; all of them together describe a bubble that has popped.

Is AI actually a bubble?

That is debated. Bulls point to fast-growing AI revenue at cloud providers and chip makers. Skeptics point to record capital spending running far ahead of the profits it produces, a pattern seen in the dot-com and railway booms. This page doesn't predict; it only reports whether the classic signs of a burst have appeared.

Where does the data come from and how often is it updated?

Stock and index prices come from Yahoo Finance (with Nasdaq.com as a fallback), Big Tech capital expenditures from quarterly SEC filings, and GPU rental prices from the Vast.ai marketplace. The page rebuilds automatically several times a day. This is not financial advice.