Compute Tightness Index

A live 0–100 gauge of how tight the AI GPU rental market is. The token-price charts everyone shares track the output of compute; this tracks the input — and the input is what moves the AI-infrastructure trade. It reads the spot-vs-on-demand discount and provider availability across H100, H200 and B200: idle capacity dumps cheap on spot (loose); scarce capacity bids spot toward on-demand and providers go dark (tight).

45.2Index (0–100)
BalancedRegime
▲ +0.430-day change
45–64Range since 2026-05-07
2026-06-24Latest snapshot

← looser  ·  45 balanced  ·  60 tight  ·  tighter →

As of 2026-06-24 the blended index sits at 45.2 (Balanced) and has been flat over the past 30 days (+0.4 pts). Listing-weighted across H100, H200 and B200 by on-demand market depth.

⤓ Download full daily series (CSV)
Blended CTIB200H200H100MI355X (thin)

By GPU — latest snapshot

GPUOn-demand $/hrSpot $/hrSpot discountListingsCTI30d
B200$6.55$3.7543%4245.8▼ -3.8
H200$4.11$2.9927%6958.2▲ +3.2
H100$3.60$1.7651%15139.1▲ +0.6
MI355Xthin · excl. from blend$8.60$0.00100%2

Spot discount = how far below on-demand the spot rate trades. A shrinking discount (rising CTI) means spare capacity is drying up; a widening discount (falling CTI) means idle GPUs are being dumped on spot. MI355X is tracked for coverage but excluded from the blended index — its listing count is too thin to weight reliably.

Methodology

For each GPU on each day: CTI = 100 × ( 0.8 · spot_avg/on_demand_avg + 0.2 · unavailable/(unavailable + on_demand_listings) ). The scarcity term is the share of listed capacity that has gone dark (clamped to 1). The blended index is the listing-weighted mean across B200, H200 and H100 (weighted by on-demand listing count, a proxy for market depth). Inputs are the daily GetDeploying aggregate of public GPU rental listings — on-demand, reserved and spot averages plus availability — one snapshot per day.

Reading it: <45 loose · 45–60 balanced · 60–80 tight · >80 acute scarcity. The spot-premium term dominates day to day; the availability term only bites when providers actually go dark, so it spikes the index during genuine capacity crunches. Young series (data since 2026-05-07); thresholds will calibrate as history accrues. Full daily history is available as CSV.

Want the underlying rows or to query this from an agent? Grab the CSV, or use the SwiftAlerts MCP endpoint (get_spot_pricing · get_inference_economics). See also the research notes and the Congress trades tracker.