A dashboard tells you a number moved. The Console tells you what moved it.

Silicon, compute, power, climate, freight and quantum in one interface, so a question that crosses all six is a query instead of a week of reconciliation.

The feeds were never the hard part.

Physical-economy data is almost always sold as separate feeds: a freight feed, a weather feed, a power feed, a hardware tracker. A feed can tell you that a vessel sailed and that a substation filed for an upgrade. It cannot tell you that the vessel is carrying the transformer the substation is waiting for, or that a campus energization date moves when either one slips. That gap is where the decision actually happens, and it is the part no feed holds.

Nuclir stores the relationships rather than the rows. A facility connects to the capacity it holds, the power it draws, the equipment on order, the routes carrying it, and the weather sitting over those routes. "What is constraining this build" becomes a path through those connections instead of four browser tabs and a judgement call.

It is also why the answer moves without anyone asking again. When a shipment clears, a queue position changes, or a storm track shifts, one relationship changes and every question routed through it answers differently the next time it is asked - in the Console and over the API at the same moment.

Context is the relationship, not a note on the record.

A number without its dependencies is a number you cannot act on. Port dwell time is trivia until it is attached to the specific equipment a specific site is waiting for; grid headroom is trivia until it is attached to the load that has contracted for it. Stored as separate metrics, both are dashboards nobody opens twice.

Here the dependency is the relationship itself, which means a query about a site returns the constraints that actually bind it. It also means the Console can say why: the path from the delay to the cause runs through the entity resolution that decided a manifest and a filing describe the same shipment, and that path is what an analyst checks before anyone reprices anything.

Four hops, asked in a single query.

This is the question the intelligence layer exists for. Answering it from separate feeds means four searches and a judgement call joining them; here it is one path, and each hop is a relationship the model already holds.

Query4 hop path

? What is constraining this build?

  1. Site
    waiting on
  2. Equipment
    shipped by
  3. Route
    exposed to
  4. Weather
Every hop is a relationship the model already holds, which is why the question can be asked once rather than four times.

Six kinds of thing, and every connection between them.

Each of these is an entity type, not a column. An entity carries its own properties and its own sources, and the relationships it holds are what the other five are reached through.

Silicon

Foundry allocation, memory supply and packaging throughput, connected to the hardware they become.

Compute

Announced and under-construction capacity, tied to the power it holds and the equipment it is waiting on.

Power

Generation, headroom, queue position and fuel, matched to the load contracted against them.

Climate

Storm tracks, heat stress, hydrology and drought, resolved onto the assets and routes they touch.

Movement

Vessels, ports, air freight and industrial flow - where the physical economy is actually going.

Quantum

Error-correction milestones, logical qubit counts, and the migration to post-quantum standards.

Public record, licensed data, and what the physical world emits.

Three source types feed the model: regulatory, market and grid data that is public, commercial datasets we license, and sensor and orbital observation. None of them agree on how an asset is named, so none of them can be joined by matching text - that is the job entity resolution does before anything reaches the layer.

Nothing is hand-curated into permanence. Every entity and relationship keeps the source it came from and when that source last said so, which is what makes the answer auditable rather than merely current. If you cannot see where a value came from, you cannot put capital or a schedule behind it, and we would rather show a thin record than a confident one. The same provenance is what disruption monitoring watches for change.

Evidence that keeps pace with the decision.

Evaluate Nuclir against the systems, markets, and decisions that matter to your organization. The result is current intelligence with the context needed to use it responsibly.