Can Northstar's map hold ten million residents?

Does the map have room for ten million residents once jobs, roads and services need space as well? An area calculation dated 19 October 2026 gives a conditional answer. A dense build-out profile reaches about 7.1 million residents; a profile built around offices and foreign trade reaches about 9.3 million. Land reclamation would add room on paper.

These are model values, not populations the city has reached. Whether such a city can be occupied, supplied and run at sufficient simulation speed is not proven.

How much area is there?

The whole map has 529 tiles and roughly 205.5 square kilometres. The complete tile inventory lists about 157 square kilometres of it as buildable land. That figure includes areas that still have to be bought, so it describes neither land already owned nor land that can be built on immediately.

A separate survey used a 200-metre grid of about 5,200 points and found roughly 48 square kilometres of water. Grid cells do not follow the map boundary exactly, so grid areas should not be added to the more precise tile areas as if they gave an exact total.

The split of the land matters more than its size. Ten million people need homes and jobs, and both need roads, utilities and public facilities. A model that counts only residential plots would ignore that competition for space.

Three build-out profiles

The calculation combines residential and job plots, road shares, larger services and a discount for target density that is not fully reached. Residential density, building maturity, residents per dwelling, jobs per resident, job mix and service footprints are varied. Each of the two bands comes from 20,000 draws from specified triangular distributions and spans the 5th to the 95th percentile. It is not an empirically confirmed confidence interval and not a full upper or lower limit of what is possible.

The fixed profile is a calculation variant; it does not mean the city could house 5.38 million today. In the office profile, about seven percent of the draws reach at least ten million. That is not a seven percent real chance of success. It shows that, under these assumptions, ten million already sits at the favourable edge of the area model.

Why offices can save space

In the zone analysis, extrapolated from an occupancy sample, about three quarters of occupied zone jobs are in office towers and about a fifth in industry. This is an estimate for those zone jobs, not the full job counter including all services. Office towers hold far more occupied jobs per zone cell than industry, so industrial jobs need far more plot cells per job; the exact values are in the second table, further below.

For ten million residents, the gross work areas including access come to about 70 square kilometres in the office profile, against just over 100 with the comparison mix. These are not building footprints.

Offices do not fill themselves. The strategy needs suitable workers, education and short routes; a job that stays empty for lack of qualification, or creates long commutes, saves little.

Foreign trade: a strategy, not a throughput result

The stored resource snapshot shows large gaps between local production and consumption, as displayed daily rates. Grain production is far below consumption, while food production is far above it; the rates are in the second table. These are snapshots of daily rates, not totals counted over a finished day. A production surplus does not show that the same amount was exported, and a deficit does not count import trips.

The planned direction is to bring area-intensive goods in from outside while office jobs grow at well-connected sites. The build log records this as the future direction chosen by the user. The existing industry has not been converted across the board. Freight needs rail and ship links, terminals and local distribution. No CPU saving and no guaranteed fall in truck trips was measured with this snapshot.

The lake as extra reserve

The grid points to a mostly deep lake: most of the roughly 48 square kilometres are deeper than 40 metres, and only about 8 square kilometres are 20 metres deep or less. A large reserve would therefore have to include deep water.

The reclamation model assumes three metres of freeboard above the water level and treats 85 to 98 percent of the new area as usable. Its conditional results for the office profile are the lower rows of the first table.

In the large partial infill, 20 to 45 percent of the water area is left free as a flat allowance for a shipping channel and pump shores. That is not a geographic proof that a particular channel stays continuous or that suitable shores exist. About 97 percent of the draws in this scenario exceed ten million residents, a figure that holds only within the model assumptions.

The model puts the fill volume at roughly 2.4 cubic kilometres for the large partial infill. That is no construction or cost plan. Effects on water level, current, shores and shipping are open. A developer request for terrain changes through the automation interface is documented and was compared with the stored reply. No write function already delivered or accepted in the game follows from it. A limited test field and a check of the water effects would come before any large project.

Services and the simulation limit

A further projection scales today's per-resident demand to ten million residents, with a reserve on target capacity. The additional power generation corresponds to about 18 nuclear-plant equivalents of the model size used; that is not a decision to build them. The counts of additional pumping stations, sewage plants and schools are in the table below. The office profile assumes a clearly more compact footprint for these large facilities; if they are less compact, computed capacity falls.

The performance model is provisional. Its fps series is calculated from the change in the simulation frame counter per elapsed second, not from rendered frames. Its input data come from a band of resident numbers far below ten million, given in the table.

For higher populations the model adds an assumed throughput limit of a few million residents, and only that assumption produces the steep drop. The resulting figure of about 27 percent at ten million is a model ratio against the nominal simulation frame rate, not a measured effective game speed.

A hard limit or a reliable forecast for ten million cannot be derived from this, and the linear memory projection does not show that memory will not become a bottleneck. Finding homes is a separate question as well: more floor space does not mean waiting households find and move into it.

Status and open points

The results support a direction: use existing space better, bring work and housing closer together and aim new work areas more at offices. Whether area suffices, homes are occupied and the simulation stays fast enough has to be shown separately, in stages that can be checked. The two full area profiles and all five reclamation scenarios were recalculated independently with the same assumptions. The snapshots come from different times and are not one atomic current state. No full construction, logistics or performance proof exists yet.