Brickroad
Brickroad
IndexSignalsBlog
/Index//

Product

  • Information Frontier Agent
  • Vendor Management Agent
  • Product Tour
  • Pricing

Solutions

  • All Solutions
  • Atlas
  • Wayfinder
  • Horizon

Suppliers

  • Sell Your Data
  • Start Onboarding
  • Programmatic Data License

Community

  • Public Index
  • Signals
  • Blog

Company

  • About
  • Team
  • Careers
  • Brand

Connect

  • LinkedIn
  • X
PrivacyTermsCookies

© 2026 Brickroad

/Index/Flop/A 60-second-cadence GPU rentable-capacity and margin panel
F

A 60-second-cadence GPU rentable-capacity and margin panel

FLOPManage supplier listing

A 60-second-cadence GPU rentable-capacity and margin panel: confirmed rentable capacity by provider and GPU class normalised into H100-equivalent units via measured rather than nominal throughput, with explicit lower-bound discipline (unknown inventory is never recorded as zero, and intervals lacking an exact observation are bridged rather than filled); live price floors per class; blended rent per H100e-hour on qualified live stock; and a model-level economics layer computing what a rented GPU earns when resold as inference tokens against live API-market quotes, yielding per-model margins, margin sensitivity ranges and break-even rents. Adjacent: index-linked GPU price forward contracts used as a forward price signal.

Open buyer room →
Flop/A 60-second-cadence GPU rentable-capacity and margin panel
SampleCoverage

Sample

Brickroad customers

Sample this dataset before you buy it.

Your sourcing agent asks Flop and files the sample in your catalog — private to you.

Start 7-day trial→Nothing is charged until the trial ends.

Coverage

Universe, instruments and categories.

Industry
Financial Exchanges & DataData Processing & Outsourced Services
Instruments
equitiesderivativescommodities
Categories
AlternativePriceSupply Chain
Regions
US
Sample tickers
NVDAAMDCRWVNBISIRENAPLDSMCIGOOGLMSFT
Dataset card
Type
not stated
Format
Tabular — live structured panels with UTC sweep timestamps and per-model/per-provider tables (no documented bulk export or public API)
Volume
Live sweep every 60 seconds: 4,503 physical GPUs measured as a confirmed lower bound, expressed as 1,391.23 H100-equivalent units of which 878.39 are priced (63.14% priced-throughput coverage), plus a per-model decision table spanning at least 7 models each carrying OpenRouter/AIHubMix/SiliconFlow quotes, benchmarked rental cost, G(n) delivery versus calibration, margin range, base margin and break-even rent, and provider-level capacity and price-floor records. At 60s cadence this is roughly 52K sweeps per provider-set per year — two orders of magnitude denser than the daily/hourly indices in this category.
Users
No registration surface or account base is described — a public read-only market monitor maintained by a 3-person team, so consumption is anonymous page loads rather than registered accounts.
History
not stated
Update frequency
not stated
Growth
Active — sweep timestamped the current day at 60-second cadence, with collectors at visibly different maturity stages (margin history begins only once both collectors hold a snapshot; some provider units flagged unvalidated), indicating an in-flight build being extended provider-by-provider rather than a static dashboard
Launched
est. 2025-2026 (no launch date published). Bounded by collectors still described as building their initial snapshot history and a 3-person organisation record with no funding history — consistent with a recent formation.
Delivery
not stated
Entity mapping
not stated
Sample
not stated
Point-in-time
not stated
Licence
not stated

supplier index read from their site

Flop has not added their own details yet. Not yet on file:

  • Dataset
  • Data dictionary
  • Coverage
  • Provenance
  • Rights & data handling