One Big Factory or Separate Ones in Satisfactory?

This is the question people ask around the point where their starter base stops being adequate, and the honest answer is that both work. What differs is which set of problems you would rather have.

The two approaches

One big factory puts everything in one place. Ore comes in at one end, finished parts come out at the other, and intermediate products never travel far.

Separate factories put each production chain near the resources it needs. An aluminium site by the bauxite, a steel site by the coal and iron, and so on, with finished parts shipped to wherever they are consumed.

What one big factory buys you

Short internal logistics. Intermediates move metres rather than kilometres. No trains to signal, no vehicle routes to maintain.

One power grid to reason about. Everything is on the same network in the same place.

Everything is visible. When something stops, you can walk to it.

No transport overhead. A train station is 50 MW plus 50 MW per platform, constantly. Ten sites means ten sets of that cost before anything moves.

What it costs you

Ore has to come to you, and ore is the worst thing to transport. It is the highest item count in any chain. Smelting at the source and shipping ingots roughly halves what you move; making plates halves it again. A central factory throws that saving away by design.

Simulation load concentrates. Everything running in one place is the worst case for frame rate, and it is felt most on console and on modest hardware. Distance between sites is genuinely free performance because the game only simulates in detail what is loaded around you.

It runs out of room. Not immediately, but a mature factory covering every production chain is enormous, and extending a packed layout is much harder than starting a new site.

What separate factories buy you

Short ore hauls. Each site sits on its resources. You move refined output, which is fewer items.

Independent failure. A stalled aluminium site does not take steel down with it.

Performance headroom. Load spreads across the map.

Room to grow. Each site has open ground around it.

What they cost you

Transport infrastructure. Trains, or long belts, or vehicle routes — plus the constant power draw of the stations at both ends.

More grids to manage, or a long power line network to unify them.

Travel time. Fixing something means going there. Hypertubes and vehicles help, but the trip is real.

Batch delivery. Trains arrive in bursts rather than continuously, so consuming sites need buffers.

The pattern most playthroughs land on

Neither extreme, and it emerges naturally rather than being planned:

Extraction and first-stage processing at the resource. Miners feeding smelters, shipping ingots rather than ore. This is where the biggest logistics saving is, and it is cheap to do.

Manufacturing centralised. Assemblers and manufacturers gathered where their several inputs converge, because those chains need many different parts and moving all of them to several places is worse than moving them to one.

Power distributed but connected. Generators near their fuel, all on one grid so surplus anywhere covers demand anywhere.

That shape gets you the ore-hauling savings without the overhead of shipping every intermediate between sites.

Deciding for a specific chain

A rule that resolves most individual cases: move the item furthest along the chain that still reduces count.

Iron ore to iron ingot is one to one, so smelting at the source does not reduce what you move — but it does mean you ship ingots, which stack better in a freight car. Ingots to plates is three to two, so making plates at the source reduces belt load by a third. Plates to reinforced plates reduces it much further.

Push processing toward the source until the next step needs an input that comes from somewhere else entirely. That is the natural boundary of a site.

Console and lower-spec hardware

Worth stating separately because it changes the recommendation. On a Series S, or a modest PC, one enormous factory is the layout most likely to become uncomfortable at a hundred hours.

Several moderate sites separated by distance genuinely help, because the console only pays the simulation cost of what is loaded around you. This is good design anyway; it is just that better hardware hides the cost of ignoring it for longer.

When to split, in practice

Rather than deciding the whole architecture upfront, split when a specific trigger appears. Three reliable ones:

The ore haul got long. If you are belting raw resource more than a few hundred metres into your main base, that chain wants its own site. Move the extraction and first-stage processing to the node and ship the output.

You ran out of flat ground. Extending onto awkward terrain costs more foundation work than starting somewhere flat. When the next expansion needs serious groundwork, that is the signal.

A new resource is far away. Bauxite, uranium and nitrogen are often nowhere near where you started. Building those chains locally and shipping the products is almost always right, because the alternative is a very long haul of the raw material.

None of these require planning ahead. They announce themselves.

Power across sites

If you go multi-site, connect the grids rather than running each independently.

A single unified grid means surplus generation anywhere covers demand anywhere, and you size total generation against total load rather than against each site’s peak separately. That is substantially cheaper, because peaks at different sites rarely coincide.

The cost is a power line run between sites, which is trivial compared to duplicating generation. Power lines are cheap, they need no maintenance, and they can follow the same route as your train track or vehicle path.

The one caution is that a unified grid fails together. A trip anywhere takes everything down. Building generation with headroom matters more on a connected grid than on isolated ones.

Common mistakes

Hauling raw ore across the map. The highest item count in the chain, moved the furthest distance.

Building ten sites early. Transport overhead outweighs the benefit until volumes are large. Start centralised and split when a chain outgrows the site.

Splitting a chain across sites. Shipping half-finished intermediates back and forth is the worst of both approaches. Split at natural boundaries, not in the middle of a chain.

Ignoring station power. 50 MW per station plus 50 per platform, constantly, whether or not a train is present.