Trade the economics of AI compute. Performance across compute inputs, capacity, inference demand and event hedges—before you express a trade.
Compute proxy · 3M — 0/7 legs tracked
Accelerators · 3M — 0/4 markets tracked
Memory / HBM · 3M — 0/6 markets tracked
GPU Clouds · 3M — 0/2 markets tracked
Perp volume · 24h $0 Hyperliquid notional · tracked markets
OF-H100 · REPLICATING PORTFOLIO H100 Compute Unit Long the scarcity and rental economics of one H100-equivalent GPU-hour.
PROXY · LIVE 3M performance — —
Maximum drawdown — Indicative daily basket
Tracking coverage 0/7 Daily market series
N 14.29%
C 14.29%
N 14.29%
D 14.29%
S 14.28%
A 14.28%
M 14.28%
01 · MARKET MAP Compute buckets See which layer of the AI production stack is moving.
Trade the full stack 02 · TRADEABLE EXPRESSIONS Basket performance Each product has its own indicative unit NAV and divisor. Base 100 is used only inside the comparable performance charts.
Create your basket Basket Unit NAV 3M chart 3M return Drawdown 24h perp volume
H1 H100 Compute Unit Tradable proxy for the scarcity and rental economics of one H100-equivalent GPU-hour.
$118.40 History loading
— — — 0/7 series tracked
Trade ME Memory Complex Equal-notional basket of HBM, DRAM and NAND producers.
$62.75 History loading
— — — 0/6 series tracked
Trade SI Accelerator Stack GPU, custom silicon and data-center networking exposure.
$342.10 History loading
— — — 0/4 series tracked
Trade CL Compute Capacity Pure-play listed GPU cloud capacity providers.
$55.20 History loading
— — — 0/2 series tracked
Trade IN Inference Assets Tokenized private inference and usage-linked capacity.
$29.65 History loading
— — — 0/2 series tracked
Trade BA Inference / Compute Basis Long tokenized inference while shorting memory input costs.
$86.30 History loading
— — — 0/6 series tracked
Trade ST Full AI Stack One coordinated position across hardware, cloud, inference and outcomes.
$214.80 History loading
— — — 0/5 series tracked
Trade 03 · PORTFOLIO PROTECTION Hedge performance Hedges are defined by the risk they offset—not by a ticker somebody happens to like.
LONG SCARCITY
Compute-cost hedge Protects: AI builders against rising GPU-hour costs
+ NVDA + CRWV + NBIS + DRAM + SKHX
SHORT CAPACITY
Demand slowdown Protects: Compute suppliers against falling utilization
− CRWV − NBIS
LONG MEMORY
HBM supply squeeze Protects: GPU buyers against memory inflation
+ DRAM + MU + SKHX + SMSN
SHORT SILICON
Chip-policy shock Protects: AI portfolios against export-control tightening
− NVDA − AMD − AVGO − MRVL
LONG OUTPUT / SHORT INPUT
Inference-margin spread Protects: Inference operators against input-cost expansion
+ VVV + DIEM − DRAM − MU
04 · INDUSTRY EVENT RISK AI industry prediction markets OpenAI is one company. This feed spans model labs, chips, capital markets, compute and policy.
Explore all predictions All
Company All Minimum horizon All 30D+ 3M+ 6M+ 1Y+ Sort 24h volume Ending soon Probability Showing active, longer-duration markets by default.
Performance methodology Comparable liquid legs are equal-notional by default. Only demonstrably thin onchain assets and bounded event premiums are capped; spread baskets remain balanced 50/50 by sleeve. Charts normalize each leg to 100 and unit NAV uses each product's own inception NAV and divisor. Values are indicative, not exchange settlement prices, and exclude fees, funding, slippage and rebalancing.