Mindustry v8: Serpulo build order and production ratios

Mindustry is a factory game where almost every number that matters is a ratio: how many drills feed a press, how much coal buys a kilowatt, how many seconds of ammo a turret holds. This guide gives the numbers for Serpulo on v8 Build 160 — the build order for the first sector, the exact consumption and production rates of every early chain, the sector unlock order, and the traps that waste a new player's first ten hours.

Every figure below was read out of the game's own content files at release tag v160.5 (core/src/mindustry/content/Blocks.java, Items.java, Drill.java, Conveyor.java, Planets.java, LandingPad.java) or from the community wiki, and the source of each is named. Where the wiki and the code disagree, the code wins and the disagreement is flagged.

What version this is for

• Game version: v8 Build 160.5, released 20 September 2026 (build 160 itself landed 11 September 2026). This is the last feature release of v8 — the developer has stated further v8 work is bugfixes only, with focus moving to v9.
• What 160 changed that matters to a builder: the copper Conveyor was re-tuned twice in two builds — 4.2 items/sec in v8.0, buffed to 6.5 in build 159, then set to 5 items/sec in build 160. Most wiki and YouTube figures still quote the old 4.2.
• Serpulo is the recommended starting planet. Erekir has a separate campaign (The Onset → Origin) with ducts and payloads instead of conveyor belts, and its campaign rework has not happened yet.

The campaign in order, and what each sector actually unlocks

Serpulo's tech tree is gated by sectors, not by research points: capturing a sector makes specific blocks researchable, and some sectors additionally require you to have researched particular blocks first. Both columns below are generated from the game's own tech-tree data. Threat levels are the game's.

Practical reading of that table: you do not need to capture sectors in map order, only in dependency order. The tech tree tells you what you are missing. Two consequences catch people out — Frozen Forest is gated behind Junction and Router of all things, and The Craters is gated behind Mender and Combustion Generator, so a player who never builds a mender is stuck staring at a locked sector with no explanation.

Also worth knowing before you plan: research spends real items, drawn from the cores of all your sectors at once. Unlocking an expensive block can empty the core you were using to build with, mid-sector.

Ground Zero: the exact opening, in the game's own order

Ground Zero is the tutorial sector, and its objectives are the build order. The enemy waves do not start until the tutorial finishes, so the timer pressure before wave 3 is entirely self-inflicted — use the free time to stockpile.

Objectives, in order:

1. Obtain 15 Copper (hand-mine it with your starting unit).
2. Build 1 Mechanical Drill on copper.
3. Move into Core: 5 Copper.
4. Move into Core: 100 Copper.
5. Move into Core: 25 Lead — completing this opens the rest of the map.
6. Build 2 Duo turrets.
7. Supply a turret (the turrets need an ammo line, not just to exist).
8. Build 6 Copper Walls. This completes and points the camera at the enemy drop zone — anything built inside that zone is destroyed.
9. Destroy 1 unit.
10. Build 1 Scatter and supply it (lead or scrap both work). The tutorial ends; wave 3 starts about a minute later.

Wave facts for the sector: about 2 minutes 10 seconds between waves, 9 waves total. Daggers every wave (1 at wave 1, +1 every other wave); Flares on waves 2, 4, 6 and 8; a single Mace on wave 8.

The Mace is the only real check. It has 550 health and 4 armour, against the Dagger's 150 health and 0 armour, and its flame weapon pierces a single layer of wall. Armour reduces the Duo's 9 damage to 5, and your starting unit's Alpha hits for 11 → 7. Build three to four layers of copper walls in front of everything, keep a copper stockpile for ammo, and wave 8 stops being a problem.

Sector supply lines for the first two hours:

• Ground Zero gives copper, lead, scrap and water, and nothing else. Capture it, then stay a while and mine out the veins to pay for the early tech.
• Frozen Forest adds coal — the sector's own advice is to find the two nearest coal veins and dedicate one to Graphite Presses and one to Combustion Generators as soon as you land. It also introduces you to the priority order that holds for the rest of the planet: stabilise defence, then power, then expand production.
• The Craters has tighter veins (especially coal) and is where Metaglass arrives via the Kiln. It is also the first sector with a guardian — a Zenith on the last wave.

Sizing drills: what a drill actually pulls per second

Drills are easier to size than the game's UI makes them look. From Drill.java, a drill's rate is:

items/sec = 60 × (ore tiles covered by the drill) ÷ (drillTime + 50 × ore hardness)

Ore hardness comes from Items.java: sand and scrap 0, copper and lead 1, coal 2, titanium 3, thorium 4. A drill can only mine ore whose hardness is ≤ its tier, which is why the Mechanical Drill will not touch titanium no matter what.

These assume the drill covers its full footprint with ore and is not boosted. Two caveats worth stating rather than hiding: the block info screen's "drill speed" figure is 60 ÷ drillTime × tiles, which ignores ore hardness and so reads slightly optimistic — the table above is the harder, correct number for copper and lead. And supplying water to a drill is listed in-game as a 256% speed boost, while the code computes the multiplier as lerp(1, liquidBoostIntensity, …) with liquidBoostIntensity = 1.6 — a factor of 1.6. The two disagree and I could not resolve which the shipped build does without running it; treat the water boost as somewhere between 1.6× and 2.56×.

The one time the formula paid off in this run: the wiki's own Craters strategy text says two Kilns need 4 sand/sec, "achievable using 7 Pneumatic Drills or 10 Mechanical Drills". The table above gives 4 ÷ 0.60 = 6.7 and 4 ÷ 0.40 = 10.0. The wiki's hand-written prose and the game's code agree, which is a good sign for both.

The crafting chains, with the numbers

Everything below is derived from the block definitions: craft time in ticks (60 ticks = 1 second), multiplied out to per-second rates.

The two that need a footnote:

• Separator. It does not produce all four outputs per craft. Separator.java picks a single result per craft with a weighted random roll: copper 5, lead 3, graphite 2, titanium 2 out of 12. So at 4 slag/sec in (1.71 crafts/sec) the expected yield is copper 0.71/s, lead 0.43/s, graphite 0.29/s, titanium 0.29/s. If you build a separator bank expecting a steady copper line, you get two in five.
• The Multi-Press is the ratio breakpoint. The Graphite Press converts 2 coal into 1 graphite with no power. The Multi-Press converts 3 coal into 2 graphite — that is 1.5 coal per graphite instead of 2, at 108 power and a large item cost. It is a coal-efficiency upgrade, not a throughput one, and 6 coal/sec is a serious belt to feed.

Drills needed to feed one machine, using the table above (full ore coverage, unboosted):

Put bluntly: a silicon smelter is a mining project. Feeding one needs about twelve mechanical drills if you are drilling the sand yourself, and sand is the input everyone forgets. If you would rather not run a sand mine, the Pulverizer turns scrap into sand at 1.5/sec for 30 power, and scrap is lying around in ruins.

Belts: how many machines one line actually feeds

So one copper conveyor moves 5 items/sec — that is three graphite presses and not a fourth (3 × 1.33 = 4.0 coal/sec, the fourth would need 5.33). A titanium conveyor feeds seven. Metaglass from a single Kiln is 2/sec, so one belt carries four kilns' output.

The side-input rule is the one that quietly ruins factories: conveyors and junctions accept items from the sides, so a belt running past a factory will happily accept that factory's output, and a belt running past a drill will feed off it. Where a line passes a block you did not mean to connect, use a Junction — it refuses side input.

Power: what a fuel item actually buys

Source values are in the game's own units (powerProduction × 60 = the number shown in-game and on the wiki).

The step change is the Steam Generator: four times the power per coal for the price of a water line. On coal alone, one Combustion Generator runs exactly one Silicon Smelter (60 vs 30 power, with headroom for a Mender). One Steam Generator runs eleven smelters' worth.

What the machines draw, for planning:

Batteries: the small Battery stores 4,000 power (5 copper, 20 lead) and the Large Battery 50,000 (20 titanium, 50 lead, 30 silicon). Buffers do not add generation — they only survive a brownout, and a brownout is not a graceful failure in Mindustry: everything on the grid slows down together, which is how defence lines collapse while the player is looking at a production line.

Defence: the early turrets, their ammo and what it costs

Ammo figures from the wiki's turret pages; build costs cross-checked against Blocks.java (they match).

Three things this table settles for the early game:

• Feed your Duos graphite, not copper. Two coal per graphite is cheap, the graphite doubles damage per shot and adds 2 tiles of range. Copper ammo is for the first ten minutes only.
• A Duo holds 15 copper or 7 graphite. An unsupplied turret is scenery, and a turret whose belt has drained at wave 6 is worse than useless because you already paid for it.
• Scatter is anti-air with no ground role, Arc and Lancer are ground with no air role. The Mace at Ground Zero wave 8 and Daggers are ground; Flares are air.

Moving between sectors: the Launch Pad, the Landing Pad, and the loadout cap

This is where v8 differs most from older guides, and where an interplanetary supply chain most often "does not work" for no visible reason.

• The Advanced Launch Pad sends, and it costs oil. Copper 350, silicon 250, lead 300, titanium 200; 480 power/sec; 9 oil/sec; 4×4. It holds exactly 100 items of one item type — it will not accept anything else until emptied — and launches a 100-item batch every 30 seconds. If it does not have 100 items when the timer expires, it waits.
• The destination needs a Landing Pad. Copper 200, graphite 100, titanium 100; 4×4. It receives 100 items per batch and burns 1,500 water per landing, so a Landing Pad at maximum rate (a batch every 5 seconds) needs 300 water/sec fed into it, and its output line needs to clear 20 items/sec. Build the receiving end first; a source pad with nowhere to send silently does nothing.
• You can only bring what half a core holds. The launch loadout is capped at launchCapacityMultiplier × core item capacity, and Serpulo's multiplier is 0.5 — so 2,000 items with the starting Core: Shard, 4,500 with Core: Foundation, 6,500 with Core: Nucleus. The dialog subtracts what your chosen schematic already needs. Only the origin sector's actual stored items count, so a sector that has been strip-mined cannot fund a loadout no matter what the cap says.
• Containers and Vaults merged into a core count as core storage. A Container adds 300 (titanium 100) and a Vault 1,000 (titanium 250, thorium 125) to the core's capacity when placed directly adjacent to it, and items inserted into them join the core's total. The wiki's Vault infobox currently says 300 — that is a copy-paste error from the Container page; the game file says 1,000.

The traps that cost the first ten hours

1. Power brownouts take the whole base down together. There is no priority system: when demand exceeds supply, turret fire rate, menders, drills and factories all slow down at once. Over-provision generation and add batteries before you add machines.
2. Research spends the items in your core, across every sector. Unlocking an expensive block mid-build can strip the copper you were using for turrets. Do research when a sector is stable, not when you are mid-expansion.
3. Sector unlocks are gated behind ordinary blocks. Frozen Forest needs Junction and Router researched; The Craters needs Mender and Combustion Generator. If a sector is locked, the missing piece is usually a dull utility block, not a weapon.
4. Everything in the enemy drop zone gets destroyed on landing. Ground Zero's tutorial tells you this rather late.
5. Enemies focus-fire the single weakest wall tile. One layer of copper wall is a shortcut straight to your turrets; the Mace's flame weapon pierces a single layer outright. Three to four layers, and repair them.
6. The Separator is a slot machine (one weighted-random output per craft, see above), and the Melter eats 6 scrap/sec to make 12 slag/sec. Scrap-based infrastructure is a throughput sink, not a free resource.
7. Sand is a recipe input you will forget. Silicon (2 sand per coal), metaglass (1 sand per lead), pyratite (2 sand) and phase fabric (10 sand) all eat it, and it is mined at only 0.40/sec per mechanical drill.
8. The Launch Pad needs oil before it can move anything, and the Landing Pad needs 1,500 water per batch. Both are easy to build and then leave idle because one input is missing.
9. A conveyor will accept items from the side — from a factory you did not mean to connect, or a drill whose output you thought was going elsewhere. Junction where lines pass blocks.
10. Overdrive Projectors do not stack. Blocks take the single highest boost available; placing three projectors over one factory wastes 420 power/sec for the effect of one.

Sources

• Game code, release tag v160.5 (byte-identical to master at time of writing) — the authoritative source for every rate, cost and capacity in this guide:
• core/src/mindustry/content/Blocks.java (build costs, craft times, input/output stacks, drill times, generator outputs, tiers, core and storage capacities)
• core/src/mindustry/content/Items.java (ore hardness), content/Planets.java (launchCapacityMultiplier = 0.5 for Serpulo)
• world/blocks/production/Drill.java (rate formula, tier rule, liquid boost), distribution/Conveyor.java and distribution/StackConveyor.java (throughput), production/Separator.java (weighted random output), world/blocks/campaign/LandingPad.java, world/blocks/power/PowerGenerator.java
• Mindustry Encyclopedia (community wiki, mindustry.miraheze.org) — read as raw wikitext via api.php?action=parse&prop=wikitext, plus the rendered Campaign table:
• Campaign (sector order, requirements and unlocks; auto-generated from the game's tech tree), Ground Zero (tutorial objective list, wave composition, wave interval), Frozen Forest, Cratered Battleground (sector strategy and the 4-sand/sec drill counts that cross-check the formula)
• Block pages: Graphite Press, Silicon Smelter, Kiln, Conveyor, Titanium Conveyor, Mechanical Drill, Pneumatic Drill, Combustion Generator, Steam Generator, Thermal Generator, Differential Generator, Thorium Reactor, Impact Reactor, RTG Generator, Solar Panel, Battery, Overdrive Projector, Power Node, Container, Vault, Launch Pad, Landing Pad, Water, Oil
• Turret pages: Duo, Scatter, Arc, Hail, Lancer, Wave, Salvo, Ripple (ammo tables, DPS and booster figures)
• Releases and patch notes: github.com/Anuken/Mindustry/releases (v160 = 11 September 2026, latest v160.5 = 20 September 2026) and the Steam release notes for v8 (14 April 2026) and build 160 (copper conveyor re-tuned to 5 items/sec; last feature release of v8).
• Reporting on the v8 release: gamingonlinux.com (15 April 2026) and the Mindustry Encyclopedia Mindustry v8 page.

Known disagreements and unverified points, stated rather than smoothed over:

• The copper Conveyor's throughput is 5 items/sec in build 160 per the game's displayedSpeed; the wiki page still lists 4.2 items/sec from v8.0.
• Drill water boost: the block-info screen and the wiki quote a 2.56× speed increase; Drill.java applies lerp(1, liquidBoostIntensity, …) with liquidBoostIntensity = 1.6. I could not determine which the shipped build does without running the game, so the drill table above is unboosted only.
• The wiki's Vault infobox lists 300 item capacity; the game file says 1,000. The Vault's build cost is titanium 250 + thorium 125 — the wiki page for it appears to be a copy of the Container page.
• Values on the wiki's auto-generated data tables were last regenerated for build 159/earlier; v160 changed only balance figures (endgame unit armour, the Nova unit, and the copper conveyor), so early-game ratios here should hold, but a mid-v9 wiki refresh may move them.