id: bulk
name: Bulk Infrastructure
domain: bulkinfrastructure.com
kind: operator
mode: hybrid
priority: 2
license: "Publicly accessible operator pages; factual data only; © Bulk Infrastructure"
attribution: "Bulk Infrastructure"
homepage: https://bulkinfrastructure.com/data-centers/locations
notes: "Campus pages /data-centers/locations/<code> carry JSON-LD LocalBusiness with postal address + GeoCoordinates; 'up to 1 GW' is recorded as planned power."
coverage: { operators: ["Bulk Infrastructure"], countries: ["NO", "DK"] }
fetch: { level: 1, maxLevel: 2, rpm: 15, concurrency: 2, respectRobots: true, maxCreditsPerRun: 20 }
discovery:
  sitemap: ["https://bulkinfrastructure.com/sitemap.xml"]
  include: ["/data-centers/locations/[a-z0-9-]+$", "/newsroom/."]
  exclude: ["\\?", "#"]
  classify:
    - { pattern: "/data-centers/locations/[a-z0-9-]+$", group: facility_pages, pageType: facility_page, priority: 60 }
    - { pattern: "/newsroom/.", group: newsroom, pageType: press_release, priority: 50 }
schedule: { facility_pages: weekly, newsroom: daily, sitemap: weekly }
extractors:
  facility: { match: ["/data-centers/locations/[a-z0-9-]+$"], kind: facility, parser: bulk_facility_v1 }
defaults: { operatorName: Bulk Infrastructure, facilityType: hyperscale }