Object
Maps
Canonical story pages with confirmed claims, competing narratives, source stacks, and update history.
About
A research publisher for AI power, not a generic AI site.
Compute Statecraft tracks how AI capability, advanced chips, capital flows, and state coalitions reshape power. The site is designed so readers and machines can audit scope, method, source quality, and revision history.
Object
Dossiers
Evergreen actor, organization, and country files with receipts, positions, networks, and timelines.
Object
Brief
A noindex daily scan that routes readers toward the canonical Maps and Dossiers instead of creating duplicate story URLs.
Publisher
Publisher of record
Compute Statecraft is the public publisher of record. Policy, corrections, and takedown routes are exposed so the chain of custody is visible.
Identity
Attribution rule
Organization-level attribution is public. Named researchers and affiliations are published only when they can be tied to verifiable public records or institutional profiles.
Audience
Decision use
The work is written for policy staff, compliance operators, journalists, and other readers who need a known source chain before acting on analysis.
Chain of custody
Institutional trust in this category depends on knowing who publishes, what standard governs publication, how errors are corrected, and where a claim sits on the spectrum from document-backed mandate to analytical projection.
Research identity rule
No human profile is published unless the identity can be tied to a public affiliation, publication record, or durable expertise trail. That rule prevents synthetic credibility from masquerading as provenance.
Current roster status
No verified individual researcher roster is published in this repository yet. The public trust surface is therefore organization-first, method-first, and correction-first.
What makes the work auditable
- Provenance labels: claims are marked SOURCED, INFERRED, or SPECULATIVE.
- Primary-source preference: official records, policy text, filings, transcripts, and direct statements are prioritized.
- Visible update history: Maps carry changelogs instead of silently rewriting assessments.
- Structured disagreement: narratives can support, contradict, or contextualize a claim without pretending all evidence is equal.
Verified mandates vs projections
- Verified mandates are anchored to official text, filings, rulemaking records, or direct statements.
- Analytical projections are treated as modelled consequences, not as legal facts.
- The boundary stays visible on the page so a reader can tell what is known from what is reasoned.
Machine-readable trust
Trust pages are intentionally structured for LLM ingestion, with stable entity relationships, site-native definitions, and page-level JSON-LD that explains what the site tracks and how it knows what it claims.
Working definitions
Hardware geofencing
A control pattern that restricts the activation, service, or performance of hardware systems based on jurisdiction, location, or approved deployment environment.
Compute integrity
The verifiable condition that a compute cluster is operating with approved hardware, software, telemetry, and policy controls intact.
Remote disablement
A hardware or platform capability that can revoke service, access, or cryptographic trust when a regulated condition is violated.
Cluster telemetry
Operational logs and measurements used to prove workload type, utilization, attestation state, topology, and policy compliance across a compute environment.
Translatory brief standard
Mandate to mechanism
State the exact clause, draft, filing, or official mandate that creates the obligation.
Describe the silicon, firmware, platform, cloud, or network surface where the obligation is enforced.
List the telemetry, attestation, procurement, or inspection path that would prove the control is real.
Policy to failure mode
Name the enforcement trigger, threshold, or prohibited condition in legal language.
Map that trigger to the system component that can throttle, deny, log, or disable behavior.
Explain what would falsify the implementation claim or reveal a compliance gap.
Operator implications
Identify which operator, vendor, or jurisdiction is bound by the obligation.
Translate the obligation into operational burdens such as monitoring, capacity planning, key management, or hardware replacement.
Show which records, audit artifacts, or public statements would confirm those burdens in practice.
What this site covers
- AI capability and model competition when compute, chips, or state action materially shape the outcome.
- Advanced semiconductors, packaging, interconnects, foundry leverage, and export controls.
- Capital allocation, procurement, subsidies, enforcement coalitions, and standards alignment.
If a topic is disconnected from those power levers, it is usually out of scope. The detailed boundary lives on the coverage page.
Read coverage prioritiesCoverage
What is in scope
The questions, sectors, and policy levers this site is built to track.
Method
How Maps are built
The claim pipeline, narrative deck, source triage, and correction loop.
Standards
How sources are treated
Primary, secondary, and social source roles, plus evidence limits.
What this site does not do
- No broad “AI news” coverage when the story lacks a compute, chip, capital, coalition, or statecraft angle.
- No truth scores, ideology meters, or automated actor ranking.
- No thin topic pages made only for search traffic.
- No claim is upgraded from inference to fact without direct source support.