The Measure Space

6. Critical Data Gaps and Research Agenda

The following gaps in our current knowledge are represent the systematic opacity that power concentration produces and maintains.

6.1 Immediate Priority Research Needs

1.Compute sovereignty ratios by nation: No public database currently measures the proportion of a given nation's compute capacity that is domestically publicly owned versus privately foreign-owned. This is the single most important metric for the democratic sovereignty argument and it does not yet exist.

2.Pension fund and sovereign wealth fund exposure audit: The link between ordinary people's savings and AI infrastructure concentration is real but unmapped. A comprehensive analysis of SWF and pension fund holdings across the top 100 AI infrastructure entities is urgently needed.

3.Global South compute dependency mapping: Existing research is overwhelmingly US and EU-centric. Original research partnerships with institutions in Africa, Latin America, South and Southeast Asia are required to map the dependency relationships that are least visible from the center.

4.Board interlock and cross-ownership mapping: Informal coordination through board memberships, investment relationships, and personal networks among AI infrastructure leaders is poorly documented. A systematic network analysis of the top 50 actors is required.

5.Military and intelligence AI contract mapping: The extent of private AI infrastructure dependency in military and intelligence applications is largely classified, creating a governance void that is among the most dangerous aspects of current concentration. Freedom of Information strategies across multiple jurisdictions are required.

6.Algorithmic governance incident database: A systematic collection of documented cases where AI systems have made consequential public decisions, in welfare, policing, border control, healthcare, finance, without meaningful human accountability. This does not exist in any comprehensive form.

6.2 Structural Data Limitations

Several categories of necessary data may be structurally difficult or impossible to obtain through open-source methods:

•

Classified military and intelligence contracts: will require either legislative disclosure mandates or insider disclosure to access

•

Real-time compute utilization: hyperscalers treat capacity utilization as proprietary; regulatory compulsion or technical inference methods may be required

•

Informal network relationships: personal and informal coordination among AI leaders cannot be fully captured through formal filing analysis

•

Futures and derivatives exposure: financial instruments giving effective control over AI infrastructure without legal ownership are poorly tracked

The project will document these structural limitations transparently and advocate for the regulatory disclosure requirements that would close them.