According to the Yardeni Research note, the experimental Claude-built tool suggests that while an AI bubble exists, the credit and interest-rate pressures associated with previous market downturns have not fully emerged.

  • Yardeni Research developed the AI Bubble Meter using Anthropic's Claude, tracking four bubble-condition indicators and two potential bear-market triggers.
  • Historically, comparable conditions produced a median inflation-adjusted S&P 500 gain of 18.3% over two years.
  • Citadel founder Ken Griffin said there are “almost certainly bubbles” in the AI boom, drawing parallels with the railroad and dot-com eras.

Yardeni Research's new artificial intelligence (AI) Bubble Meter suggests an AI stock market bubble exists, but the forces that historically ended major market booms have not fully emerged.

In its Wednesday briefing, the research firm said its experimental tool, developed using Anthropic's Claude Fable 5.1 model, registered a “High” reading of 83 out of 100 for bubble conditions, compared with a “Low” reading of 34 for potential bear-market triggers.

Yardeni said the tool, which remains a work in progress, evaluates whether conditions resemble previous bubbles and whether the forces that ended those booms are developing.

What History Says About AI Bubble Risks?

Yardeni's analysis identified 54 quarters since 1965 when bubble conditions were high, but trigger readings remained low. The S&P 500 entered a bear market within the following two years in only 9% of those cases.

The index's median inflation-adjusted return over the following two years was an 18.3% gain. When both indicators registered high readings, the historical frequency of a bear market within two years climbed to 42%, with a median loss of 12.5%.

The firm said the historical findings suggest the current combination of readings is unlikely to precede a bear market over the next two years, although risks would increase if the Triggers dial moved toward a high reading.

On Stocktwits, retail sentiment toward the SPDR S&P 500 ETF Trust (SPY), which tracks the benchmark index, was ‘bearish’ at the time of writing.

What Could Trigger An AI Bubble Burst?

The Claude-assisted model tracks six indicators: four measuring bubble conditions, including valuations, household equity exposure, market returns and private investment, and two tracking potential triggers through credit growth and Treasury yields.

Yardeni's backtesting found that credit and interest rates, rather than valuations, were more effective at identifying the timing of previous market downturns. The firm also noted that the two-year Treasury yield had risen 115 basis points from its December low, while credit remained relatively unstretched outside AI-related sectors.

Yardeni is monitoring whether the two-year Treasury yield continues moving toward the 200-to-350-basis-point increases that preceded the 2000 and 2006 bubbles. It is also watching for signs of idle capacity in AI hyperscalers' third-quarter earnings reports due at the end of October.

Retail investors’ sentiment toward the iShares 1-3 Year Treasury Bond ETF (SHY), which tracks short-term U.S. Treasury bonds, was ‘neutral’ on Stocktwits.

Ken Griffin Sees Bubbles Within AI Boom

Separately, Citadel founder Ken Griffin said Tuesday that there are “almost certainly bubbles” in the current AI boom. Speaking at Citadel Securities' Future of Global Markets conference, Griffin compared the AI investment boom with 19th-century railroad speculation and the dot-com bubble.

He noted that the internet boom produced failed businesses such as Pets.com alongside enduring companies such as Amazon (AMZN), illustrating how speculative excess can accompany major technological transformations.

See Also: SMCI Teams Up With Nokia, Nvidia To Bring AI Computing To 5G Networks — Eyes AI-Native 6G Expansion

For updates and corrections, email newsroom[at]stocktwits[dot]com<