Quantitative survey

• Base: n=701 decision-makers
• Partner: Censuswide
• Fieldwork: February 12–25, 2026
• Regions: UK, US & Europe (France, Italy, Portugal, Romania, Spain, Nordics)

Qualitative material

• Interview and external benchmark material is labelled separately from the survey and is not blended into survey percentages
• Findings triangulated against the registered source base
• Independent benchmark data (e.g. Artificial Analysis) retained and labelled as such
• Structured expert interviews (E3) are planned but not yet collected; interview counts are held for review before public display

Confidence framework

Every claim carries a confidence rating — High, Medium or Low — with a stated reason, and a review date. Of 69 tracked claims, 21 are High, 28 Medium and 20 Low confidence. A claim is only shown publicly once it is approved for display, cleared for rights, sourced, dated and within its freshness window; the remaining claims are held for review.

Shared entities

Geographies, markets, workload profiles and power sources are defined once and reused across every watch. This lets the Power & Site Intelligence Map, Energy Migration Matrix and placement logic reference the same underlying records rather than disconnected source models.

Claim types

Observed

A directly reported figure from a registered source, quoted as published.

Survey finding

A result from the CUDO Land. Power. Compute. survey, shown with its base and question.

Calculated

A figure CUDO derives from other inputs using a stated, reproducible method and a calculation receipt.

Estimate

An approximate figure where the underlying data is incomplete; the basis and range are stated.

Forecast

A forward-looking figure attributed to its source, presented as a projection, not a measurement.

CUDO interpretation

An editorial reading of the evidence, clearly attributed as interpretation rather than a measured fact.

Evidence status

Verified

Confirmed against a registered source and reviewed by a person.

Partly verified

Some elements confirmed; others still carry a stated caveat.

Calculated

Produced by a CUDO method with a reproducible receipt.

Estimate

Approximate; basis and uncertainty are stated.

Forecast

A projection attributed to its source.

CUDO interpretation

Editorial synthesis, attributed as such.

Not verified

Not yet confirmed; held back from public display.

Stale

Past its freshness window; withdrawn from display until re-checked.

Withdrawn

Removed after review; not shown publicly.

Price basis

Wholesale market price

Spot or day-ahead market price.

Retail sector average

Average sector tariff, not a single contract.

Published tariff

A rate published by a supplier or regulator.

Public list price

A vendor list or advertised rate.

Contracted price

A specific contracted rate. Not shown publicly without explicit authority.

PPA price

A power purchase agreement price. Not shown publicly without explicit authority.

Forecast

A projected price attributed to its source.

CUDO scenario

An illustrative figure built by CUDO for a stated scenario, not a market observation.

Legal instrument status

In force

Currently applicable law.

Adopted, not yet applicable

Enacted, with a future application date.

Official guidance

Regulator guidance, not primary legislation.

Legislative proposal

A proposal under consideration.

Draft scheme

A draft not yet adopted.

Policy commitment

A stated intention, not yet law.

Repealed or superseded

No longer in effect.

Requires legal review

Flagged for legal review before any public reliance.

Freshness and stale rules

• Each claim, signal and event carries a data period, a retrieved date and a last-verified date.
• Each record also carries a stale-after date. Once that date passes, the record is treated as stale and withdrawn from public display until it is re-checked.
• The named release cut-off is 15 July 2026. Figures reflect evidence available on or before that date.

Calculation receipt method

• Every published CUDO calculation (E4) shows a receipt: the formula, the input claims, the assumptions, the exclusions, the period and the rounding rule.
• A calculation is only published when it is approved, reproducible from its stated inputs, reviewed by a person and within its freshness window.
• Where an input is a scenario rather than a market observation, it is labelled as a CUDO scenario.

Correction and withdrawal policy

• If a published figure is found to be wrong, it is corrected against its registered source and the review date is updated.
• If a claim can no longer be supported, it is withdrawn and no longer shown publicly, rather than left in place with a caveat.
• Where a correction changes a headline figure, the change is reflected in the Latest Update feed.

Rights and download policy

• A source is only shown in the public register when its display rights are approved.
• Download rights are tracked separately from display rights. Where a source is not cleared for redistribution, the Observatory links to it rather than reproducing it.
• Chart exports carry CUDO attribution and the data cut-off, and reproduce only figures cleared for public display.