🧠 Technology & AI mixed · 6–18 months
A what‑if from the future

What if a memory-chip glut sends DRAM prices crashing 50 percent?

A 50% DRAM/NAND spot crash from oversupply guts Micron/SK Hynix margins — a memory-specific glut (semiconductor_risk negative), which is why fab/litho names hold or firm rather than fall. Rhymes with the 2018–19 and 2022 memory down-cycles where commodity-memory pricing collapsed while logic held up better. Forward angle: HBM for AI is the swing — if the glut is commodity DDR while HBM stays tight, Micron's blended hit is smaller than the spot print implies; watch the mix, not the headline.

28%
our model probability
over 6–18 months
prediction markets — wisdom of the crowd
loading live odds…
Empirically anchored 28% · 90% range 0–55% · 10 analogues · measured class tech_ai_bull 35% in 18 mo · 3% held back for the unknown
how we built this number — every step
Measured class rate — tech_ai_bull ≈0.2842/yr → 35% in 18 mo35%
Analyst prior · editorial share 100% of the class45%
Pooled · weight 62%29%
Crowd — no liquid market
Reserve 3% · no extremizing (×1.0)29%
Published28%

The class rate is measured from our dated, sourced event library (decade-normalized Poisson — the full table is public at base_rates.json). The variant’s share within its class is the analyst’s editorial call, published so you can audit it. A wider range means thinner precedent. Full recipe: methodology · scored at Reality Check.

The butterfly cascade

How this trigger trickles across markets, left → right — the root shock, its first‑order moves, then the ripple effects. Drag any node; tap a market for its real price history.

Resolution timeline — how this probability is moving

Our model's odds (gold) over time vs the crowd's (Polymarket, blue), from the past toward the 6–18 months horizon. Each dot is a real macro event that nudged the probability — green pushed it up, red pushed it down. Tap a dot for the source. The gold path is an illustrative reconstruction anchored to today's estimate — real dated events, not a live re-estimate history.

loading the timeline…

What it would mean

If this plays out, it is a mixed shock. Memory makers flood the market, sending DRAM and NAND spot prices down 50% and gutting Micron and SK Hynix margins. The trigger decomposes into signed root‑shocks — AI capex ▼ · Semiconductor supply risk ▼ — which propagate through our causal graph to the markets below.

If it happens — the markets it would move

Biggest moves first. Projected moves are cascade-model priors; hist A–B% = what comparable past events actually did (measured abnormal returns), and model prior · unmeasured marks markets with no analogue backing yet. Tap any market for its price history.

MarketClassProjected move
1TSMC TSMon Hyperliquid 📈 chartEquity▲ +1.1%
hist -1.46–+2.2% · other way +2.11% (n=12)
2ASML ASMLon Hyperliquid 📈 chartEquity▲ +0.5%
hist -2.33–+2.17% · other way -1.97% (n=12)
3Semiconductors SMHon Hyperliquid 📈 chartEquity▲ +0.2%
hist -0.33–+0.58% · other way +1.52% (n=12)
4Nvidia NVDAon Hyperliquid 📈 chartEquity▲ +0.2%
hist -0.79–+0.63% · other way +2.48% (n=12)
5Broadcom AVGOon Hyperliquid 📈 chartEquity▼ -0.1%
hist -1.0–+0.55% · other way +0.78% (n=12)
6AMD AMDon Hyperliquid 📈 chartEquity▲ +0.2%
hist -0.75–+0.71% · other way -1.2% (n=12)
7Marvell MRVLon Hyperliquid 📈 chartEquity▲ +0.2%
hist -0.96–+1.0% · other way +2.1% (n=12)

Probable recommendation

If the scenario above plays out, the probable cross‑asset positioning → a scenario‑conditional read, not personalized investment advice
For a common-man portfolio: Mixed for a typical portfolio — the move is more about rotation than direction. Favour the winners over the losers below rather than net exposure.

Why we may diverge from history

Trust measured history's short: AMD/TSM/NVDA analogues are clean, recent, on-channel chip-supply selloffs (ASML miss, Micron guide-down, DeepSeek) — the cascade's bullish read over-reaches against a real DRAM glut.

Historical precedent — what analogous events actually did

Across 10 analogous events (overlap‑weighted), as abnormal returns — market beta stripped, so it's the event's own effect, not the market backdrop. Shown at 20 days (persistent) and 5 days (immediate); ↺ fades = the two horizons disagree. Confidence = consistency × sample × significance.

TSMC slumps as DeepSeek roils AI-chip demand assumptions 2025-02 Megacap AI-capex doubt selloff 2024-07 Micron's weak FQ2 guidance sparks a sharp December selloff 2024-12 ASML bookings-miss crash 2024-10 Trump 'Taiwan should pay for defense' chip selloff 2024-07 Nikkei 225 surpasses its 1989 bubble peak 2024-02 Netflix subscriber-loss crash 2022-04 Meta 2022-02 Didi removed from China app stores after NYSE IPO 2021-07 Nvidia crypto-glut guidance crash 2018-11
AssetHistory saysAbnormal (20d · 5d)HitnConfidencevs cascade
Gold XAULONG+1.4% · 5d -0.5% ↺ fades73%10 0.36·
10y yield DGS10SHORT-15bp · 5d 0bp73%10 0.36·
Bitcoin BTCSHORT-4.7% · 5d -2.0%73%10 0.35·
High-yield credit HYGLONG+0.8% · 5d +0.3%73%10 0.34·
NVDA NVDASHORT-0.9% · 5d -3.7%73%10 0.29⚠ differs
AVGO AVGOSHORT-0.9% · 5d -3.2%64%10 0.22✓ matches cascade
US dollar DXYSHORT-0.2% · 5d +0.2% ↺ fades64%10 0.21·
ASML ASMLSHORT-2.9% · 5d -5.5%64%10 0.20⚠ differs
TSM TSMSHORT-2.2% · 5d -2.3%64%10 0.19⚠ differs
AMD AMDSHORT-0.9% · 5d -4.3%64%10 0.19⚠ differs
SMH SMHSHORT-0.5% · 5d -2.0%55%10 0.08⚠ differs
MRVL MRVLSHORT-1.2% · 5d -1.5%55%10 0.07⚠ differs
Volatility VIXLONG+3.2% · 5d +0.7%55%10 0.07·

Why this probability

Memory is cyclical; 50% DRAM/NAND spot crashes are common base-rate over 6-18mo, though AI demand cushions. A base‑rate‑anchored prior, continuously scored against what actually happens — not a forecast.

Methodology. Probability and impact are anchored to history and scored against what actually happens — wins and losses, in public, at Reality Check. Crowd odds live from Polymarket & Kalshi. By Vikas Singh, Quantitative Strategist. Updated 2026-07-03.