01 / The mechanism and its boundary
What the technique establishes
Remote detection of data centres is a set of methods for finding facilities and estimating their size from outside. Large AI data centres need buildings, substations and cooling equipment, and they shed roughly as much heat as the electricity they use. These features can be seen without the operator's cooperation. Analysts already combine satellite imagery with permits and utility filings to track the construction of known large facilities and estimate their power capacity. For verification, the harder task is finding facilities nobody has declared. As of September 2026 no systematic search for such facilities has been published, and automated detection of data centres remains mainly conceptual. Imagery also cannot see inside buildings or count chips. The main weaknesses are concealment, such as disguising a facility as other industry or building it underground, and sites too small to stand out. Verification frameworks treat these signals as supplements to stronger mechanisms.
- Threat model
- Adversarial prover
- Adversarial evaluation
- Published analysis
- Hardware needed
- None
- Prover cooperation
- Not required
- Confidentiality
- Preserving
- Category
- Remote & side channel sensing
Claims and scope
A direct link identifies the intended claim. A supporting link supplies part of the evidence. Neither establishes that a complete verification system has been demonstrated.
There is no undeclared relevant compute
Searches for large facilities that have not been declared.
Compute stock is at most a declared amount
Estimates the power capacity, and so roughly the compute, of observed facilities.
Readiness for a stated use
Assessed use: finding undeclared data centres above an agreed compute threshold
medium confidence · current · assessed 2026-10-08 · rubric 1.1
This is the source map’s editorial assessment. Production use is not evidence of resistance to every adversary.
R1 for finding undeclared facilities, because no public work shows a systematic search that finds them.
- R1 met: Halstead and Larsen describe heat, imagery and other detection signals, ways to conceal a facility and the odds of detecting covert ones S-1410. Baker et al. place satellite imagery among supplementary verification mechanisms S-0002.
- R2 not met for this use: Krawec's case studies track known sites S-1409. Epoch AI reports that its public database finds large facilities mainly through company announcements, news, third-party databases and social media S-3301. Krawec states that automated data-centre detection "remains primarily conceptual at present" S-1409.
The level is for the primary use (There is no undeclared relevant compute). For the supporting use of estimating the capacity of known sites (Compute stock is at most a declared amount), the public Epoch database, with its published error estimates, would meet at least R2 S-3301 S-1411.
Evidence needed for the next level
Published end-to-end results on finding previously unknown large facilities over a wide area, with measured miss and false-alarm rates.
An evaluation against a stated concealment adversary, for example disguised or underground facilities.
Limitations, flaws, and blockers
These are attributed assessments from the source map. Absence of a listed flaw is not a security guarantee.
significant / open / theoretical argument
Facilities can be disguised or hidden
Halstead and Larsen discuss two ways to hide a facility. One is to disguise it as a legitimate industrial site. The other is to build it underground, with cooling that avoids visible heat plumes. They note that the underground option requires bespoke engineering S-1410.
significant / open / theoretical argument
Small sites may not be detectable
Halstead and Larsen conclude that a sufficiently small covert project could not be ruled out with confidence. In their estimates, the chance of detection is lower for smaller sites S-1410. Krawec notes that small data centres in existing buildings may lack the distinctive features of large facilities S-1409.
significant / open / open question
Search for unknown sites is undemonstrated
Krawec reports that telling data centres apart from other industrial facilities systematically is difficult. Automating detection would need large amounts of training imagery and a purpose-trained model. In Krawec's words, automated data-centre detection "remains primarily conceptual at present" S-1409.
What still blocks use or stronger assurance
- S-1409
Wide-area, automated detection of data centres is not yet practical and needs large training datasets.
- S-1410S-1409
No measured detection or false-alarm rates for finding undeclared facilities have been published.
- S-1409
Recent high-resolution imagery is costly, is limited by weather and needs trained analysts.
Connections in the research map
Complementary techniques
Concepts used
Organizations and developers
Sources and provenance
- S-1409 / Tier B
Tracking Hyperscale AI Data Center Growth with Satellite Imagery ↗
C. Krawec · 2026 · Federation of American Scientists
Supports: observable features; imagery sources and limits; capacity estimate example; cannot see inside; automated detection conceptual and its data needs; IAEA analogy; future sensors
Locator: Methodology; Opportunities and Challenges; Case Studies 1-2; Recommendations; Opportunities for Further Research
Version and catalogue details - S-1411 / Tier C
Introducing the Frontier Data Centers Hub ↗
Epoch AI · 2025 · Epoch AI
Supports: public dataset; cooling-equipment-based capacity method; coverage figures; licence
Locator: announcement post
Version and catalogue details - S-3301 / Tier B
AI Data Centers Documentation – Methodology ↗
Epoch AI · 2026 · Epoch AI
Supports: how sites are found; reported accuracy of IT-power estimates; estimated coverage of global AI computing capacity (provider-reported)
Locator: Coverage; Cooling model; Analysis
Version and catalogue details - S-1410 / Tier C
Covert AI Projects ↗
B. Halstead, T. Larsen · 2026 · AI 2040
Supports: heat-balance argument; detection signals; confidence in locating non-concealed data centres; concealment strategies; intuition-based detection probabilities and their conditions; limits for small projects; undeployed chips
Locator: detection sections; direct observation of AI datacenters; table of intuition-based detection probabilities by site size and number of sites
Version and catalogue details - S-0002 / Tier B
Verifying International Agreements on AI: Six Layers of Verification for Rules on Large-Scale AI Development and Deployment ↗
M. Baker, G. Kulp, O. Marks, M. Brundage, L. Heim · 2025 · RAND Corporation
Supports: satellite imagery, OSINT, supplier information and financial audits as supplementary mechanisms; national intelligence layer
Locator: §4.3, §4.4
Version and catalogue details - S-0007 / Tier B
Hardware-Level Governance of AI Compute: A Feasibility Taxonomy for Regulatory Compliance and Treaty Verification ↗
S. Ansari · 2026 · arXiv
Supports: infrared imaging can detect undeclared data centres; power alone cannot separate AI from other HPC
Locator: §3.1 (M4)
Version and catalogue details - S-3300 / Tier C
Limitations of Satellite Imagery Analysis for AI-Specific Data Centers ↗
L. Heim, K. Pilz · 2024 · Lennart Heim's blog
Supports: AI data centres not visually distinguishable from other data centres (2024); AI compute housed in existing campuses
Locator: blog post
Version and catalogue details
- Source review date
- 2026-09-25
- Drafted by (source map)
- ai
- Review handles (source map)
- codex-review