AI Bubble Clock
Midnight = the top (March 2000 was the last one — Nasdaq then fell 78%). The hand is a composite of six hard signals vs that peak. Not a timing call — a read on how late it is.
⟳ Rebuilt daily · Updated: —
Composite read
≈ 9:30
2026: late, frothy · 2000: ≈ 11:40 — minutes from the top
TL;DR
CHEAPERIndex 22× fwd earnings vs 27× in 2000. Nvidia ~38× vs Cisco's ~130× at the peak — and the leaders print real FCF this time.
HEAVIERTop-10 = ~38% of the S&P 500 (was 27%). The index is the AI trade — no diversification left.
REFLEXIVE~$800B of circular deals: Nvidia funds OpenAI → pays Oracle → buys Nvidia. Demand is partly self-printed.
LEVEREDMargin debt ~3.6% of GDP vs 2.7% in 2000; retail inflows ~2× the 5-yr pace.
Bull case is intact because cash flow is real. The break risk is concentration + capex ROI, and the consensus high-risk window is 2026–2028.
The 6 signals — 2000 vs now
0–100 "froth" score per signal. Red = 2000, green = now. Green longer than red = worse today.
Where the premium sits — the AI stack
Froth is densest at the base (silicon, power, the private labs), thinnest at the app layer that has to actually monetize. Inverse of 1999, when the consumer dot-coms ran hottest and the infra lagged. Whoever sells compute wins regardless of which app survives.
100 names mapped — how "AI" is each, really?
X = proximity to the AI compute core (Nvidia ≈ 100). Y = valuation richness (blended P/E + P/S percentile, so loss-makers still plot). Bubble = market cap. Filter by group; hover for detail. Top-right = max beta; top-left = AI-washing.
← loosely tied | proximity to AI core | the engine room → · bubble = mkt cap
↗ Top-right: rich + core. NVDA, AVGO, PLTR, CRWV, ALAB — max upside, max drawdown.
↖ Top-left: rich, weak AI tie — narrative beta / AI-washing risk.
↘ Bottom-right: core but cheaper — memory, foundry, some hyperscalers. Where value sits.
↙ Bottom-left: low exposure, low premium — barely in the trade.
Full list — 100 names (click a header to sort)
Educational snapshot, not investment advice; nothing here is a recommendation. Proximity/valuation scores are hand-calibrated to public data — directional, not precise. Macro figures are 2026 consensus / press estimates and move daily.