Skip to content
CapexSignal.

Coverage · India

Two gates, not one: permission, and knowing it's the same project.

India's Central Electricity Authority publishes a quarterly renewable-energy project table as a PDF, not a machine-readable export. Our adapter is implemented and live-layout verified across 654 nationwide rows, but two separate gates — not one — sit between that and a customer-ready geography.

What CEA RE is

The Central Electricity Authority's renewable-energy report lists projects by state, technology and capacity, refreshed each quarter. It is authoritative — the CEA is India's central electricity planning body — but it is published as a formatted PDF rather than a structured file, which is a genuinely harder extraction target than a CSV or an API export.

Brazil's SIGA, by contrast, is a queryable open-data export with stable fields — which is exactly why Brazil's remaining problem is a licence question rather than an identity one, and why the two geographies are gated so differently despite both having implemented, verified adapters today.

Gate one: reproduction permission

Reformatting and redistributing a government PDF's tables commercially is a different legal question from reading a register that already offers structured open data, the way Brazil's SIGA does. CEA RE needs its own explicit reproduction permission before rows drawn from it can sit inside paid customer storage.

Gate two: durable identity across quarters

A quarterly PDF carries no stable row key. Unless a project can be recognised as the same project from one release to the next — despite reordering, renaming or reformatting in the source — a said-versus-did comparison or an append-only history is not trustworthy, because "the value changed" and "we matched the wrong row" would look identical. Solving this is an identity resolution problem specific to low-frequency, unkeyed sources, distinct from the licence question Brazil faces.

The general rule elsewhere on the platform is to merge two records only when an authority settles the question — a company registry identifier, for instance — and to send the pair to review rather than guess when no authority can. CEA RE has no equivalent per-project registry identifier to lean on, so its cross-quarter matching has to be built and tested as its own case rather than reused from a source that already carries one.

What clears before paid access

Both gates have to close, and then India follows the same remaining path as any other geography: a fresh production import, seven consecutive daily canaries, and a published accuracy scorecard. No adapter is represented as customer-ready before all of them pass, and a permission letter alone would not shortcut the identity work that still has to be built.

Questions people ask about India specifically

Why does India need two gates instead of one?

Because reproduction permission and durable identity are separate problems. Permission is a legal question about reusing CEA’s published PDF tables commercially; identity is an engineering one, because a quarterly PDF gives no stable row key, so the same project has to be recognised as itself across quarters without one.

How large is the CEA RE dataset?

The adapter is live-layout verified across 654 nationwide rows in the current quarterly release — a national count, published four times a year rather than continuously.

Why is a PDF harder to adapt than a database export?

A PDF table has no row identifiers and can reformat silently between releases, so extraction has to re-establish both the layout and which row is which project every quarter, rather than reading a stable schema once and trusting it going forward.

See United Kingdom, United States and Brazil for the other implemented adapters, or the full coverage table.