How SMRMap finds and updates records

SMRMap is a source-linked directory. This page explains how projects, designs, companies and programmes are found, checked and kept current, which sources are watched, and where coverage is still thin.

Source list reviewed 14 October 2026; country research matrix cutoff 2 October 2026.

The five-step method

1. Find. Records start from named sources: international inventories (IAEA ARIS and PRIS, the NEA SMR Dashboard and the NuclearChain export of Global Energy Monitor's tracker), national regulators, governments, developers and trade news.
2. Match. A new item is compared with the existing projects, designs, companies and programmes by name before anything is added, so one plant is not listed twice and a renamed project keeps its history.
3. Verify. A fact is published only after a reviewer has opened a source that states it. Every record lists its sources and the day it was last checked; unknown values stay unknown.
4. Maintain. Records come up for review on a fixed schedule: projects in licensing, construction or commissioning after 30 days, announced and siting projects after 60 days, other stages after 90 days, design profiles after a year and programmes and agreements after 180 days.
5. Discover. Each week a bounded set of feeds and listings is checked for new items: trade news, regulator and government listings and developer newsrooms. A new item becomes a lead for a reviewer; it is never published on its own.

Source families

A page that serves two families is counted in each and read once. Open-question watches follow a specific missing fact on an existing record, such as a pending decision or a site that has not been named yet.

Weekly checks

The worldwide trade-news feeds are read every week. The other feeds and listings take turns, oldest-checked first, on a planned cycle of 2 weeks. Each weekly run has a time limit: when slow sources use it up, the remaining sources are deferred to the next run and reported as deferred, so the real interval can be longer than planned. International inventories, and newsrooms that refuse automated requests, are reviewed by hand on a fixed cadence.

Automatic checks raise leads for review and never publish facts. A feed or page that could not be read is recorded as unread, which is not the same as "nothing new was published".

Coverage by country

Countries with at least one published project or programme. A source counts for a country only when it is a national source for that country; worldwide feeds are listed separately and never count as national coverage. 8 of 69 countries have an automatic national source.

Developers

Reactor developers whose designs are linked to at least one published project or programme. 19 of 75 have their newsroom checked automatically or reviewed on a cadence.

Detail completeness

Which expected facts are known for published records, by project stage. More facts are expected as a project advances: every project needs a design, a developer, a named site and a primary source; from licensing on, its capacity, module count, regulatory action, operator and map coordinates; from construction on, construction evidence and an operation target; and an operating project needs evidence of operation. Each cell reads known/expected. "Unknown" means not yet found in a source, not zero, and derived figures are not counted as known. A dash means the fact is not expected at that stage.

Derived plant outputs (planned modules times a single per-module rating) are never counted as known.

Design profiles

Programmes and agreements

Limits

• SMRMap is not a census: its counts describe this directory, not every small modular reactor effort in the world, and a record is only as current as its last check date.
• Worldwide feeds do not prove national completeness: a country without a national source can still have activity that no feed reported.
• Withheld names, undisclosed sites and national sources that refuse or block access remain gaps until a readable source fills them.
• Country research routes, source snapshots and open evidence questions are listed on the coverage and gaps page; definitions are in the methodology.

Dataset licence: CC BY 4.0.