About GCPI Research

An independent academic programme on the measurement of AI inference costs and related market structure.

Programme

GCPI Research

Independent research programme on the measurement of AI inference costs and related market structure.

GCPI applies event-study and risk-pricing habits to public inference pricing — a market with supply shocks, information asymmetry, and no shared spot convention.

Scope

What this site is

Public release surface for compute research indices (CN-Token Spot and related designs), methodology pointers, and open downloads. It is not a trading venue, vendor catalogue, or procurement tool.

Operational price panels are curated in ComputeNav; index construction and documentation are maintained in this research repository.

How to Cite

Citation formats

APA

GCPI Research. (2026). Global Compute Price Index (GCPI) research indices (v0.3.0) [Dataset]. https://www.globalcomputeindex.org/

BibTeX

@dataset{gcpi2026indices,
  author    = {{GCPI Research}},
  title     = {Global Compute Price Index ({GCPI}) research indices},
  year      = {2026},
  version   = {0.3.0},
  publisher = {GCPI Research},
  url       = {https://www.globalcomputeindex.org/},
  note      = {See handbook/ for series-specific methodology.}
}

A Zenodo DOI will be assigned upon the first formal release. Until then, cite methodology and data versions as above (docs/CITATION.cff).

License

Creative Commons BY 4.0

Data and methodology documentation are released under the CC BY 4.0 licence.

Independence

Editorial independence

GCPI Research has no financial relationship with covered vendors. List prices are public. Not financial, investment, or procurement advice.