planned2027
AI Infrastructure Operations Survey — 2027 Edition
A planned survey of how teams actually operate AI infrastructure in production — what breaks, what is measured, and where engineering time goes. Not yet conducted.
Research
Research
Most published numbers about AI infrastructure come from someone selling a layer of it. Our research programme exists to produce measurements with a stated method that anyone can argue with. Nothing here is published before it has been run.
How we publish research
Four commitments that decide what appears on this page and what does not.
Methodology before results
Every published report states how the measurement was taken, in what environment, with which versions, and what we could not control for.
Data is published
Where a report is based on measurement, the underlying dataset is published with it under a stated licence. A chart you cannot check is an assertion.
Nothing is announced as finished
Projects that have not been run are listed as planned. There are no preview figures, no illustrative numbers, and no placeholder findings.
Reproducible where possible
Benchmark code goes in a public repository so someone can disagree with us using the same harness.
Programme
Research projects
Current state is shown for each project. Nothing is presented as a finding until it is one.
planned
GPU Cloud Price Index — Methodology
A planned recurring measurement of accelerator pricing across cloud providers, normalised so the comparison is meaningful. Methodology published before data collection begins.
Newsletter
Inside The AI Stack Brief
A practical weekly briefing on AI engineering, infrastructure, production operations, and the technologies powering the AI stack.
One email a week. No sponsorship placements inside the technical sections. Unsubscribe in one click.