Skip to content
DraconiticResearch
Search Sources API
203,952 studies31 sourcesUpdated 58 min ago
PreprintCC BY 4.0Zenodo

Sovereign Inference Economics: The Real Crossover Point Between On-Premise and Cloud Inference

  1. 1Mickai LTD

Abstract

A technical paper setting out the five variables (hardware amortisation, power draw, utilisation, request volume, model size) that decide whether on-premise AI inference is cheaper than a cloud API, with a worked illustrative example showing how the crossover point shifts with request volume.

Related

  1. Anytime-Valid Inference in Linear Models with Applications to Regression-Adjusted Causal Inference
    Michael Lindon, Dae Woong Ham, Martin P. Tingley +1
    Journal of the American Statistical AssociationPeer-reviewedLicence not statedFull text15 citedStatisticsOpenAlex
  2. Quantitative gene-level inference of tumor transcription from plasma epigenomes
    Research SquarePreprintCC BYFull textEurope PMC
  3. Semiparametrically efficient semi-supervised statistical inference
    Zichun Xu, Daniela Witten, Ali Shojaie
    BiometrikaPeer-reviewedLicence not statedFull text32 citedStatisticsOpenAlex
  4. Adaptive Conformal Inference through the Lens of Blackwell Approachability
    Guillaume Principato, Gilles Stoltz
    HALPreprintLicence not statedStatisticsHAL
  5. Correcting for ascertainment bias in the inference of population structure
    Gilles Guillot, Matthieu Foll
    BioinformaticsPeer-reviewedLicence not statedFull text34 citedGenetics & genomicsHAL
BibTeX
@misc{irons2026sovereign,
  title = {{Sovereign Inference Economics: The Real Crossover Point Between On-Premise and Cloud Inference}},
  author = {Micky Irons},
  howpublished = {Zenodo},
  year = {2026},
  month = oct,
  doi = {10.5281/zenodo.23512629},
  url = {https://zenodo.org/records/23512629},
  note = {CC BY 4.0},
}
RIS
TY  - UNPB
TI  - Sovereign Inference Economics: The Real Crossover Point Between On-Premise and Cloud Inference
AU  - Micky Irons
PY  - 2026
DA  - 2026/10/20
PB  - Zenodo
DO  - 10.5281/zenodo.23512629
UR  - https://zenodo.org/records/23512629
AB  - A technical paper setting out the five variables (hardware amortisation, power draw, utilisation, request volume, model size) that decide whether on-premise AI inference is cheaper than a cloud API, with a worked illustrative example showing how the crossover point shifts with request volume.
ER  -