Last updated: July 4, 2026
shapez 2 MAM Guide
AMake Anything Machine (MAM) is the ultimate expression of automation in shapez 2. While a normal factory produces one specific shape repeatedly, a MAM reads the Hub delivery goal and automatically reconfigures itself to produce whatever shape is requested. Building your first MAM is a rite of passage — it forces you to understand every game mechanic at a deep level. This guide walks you through MAM components, the step-by-step build order, fluid throughput optimization, and the most common mistakes that cause MAMs to fail silently.
What Is a MAM and Why Build One
A MAM (Make Anything Machine) is an automated factory that can produce any shape the Hub requests without manual reconfiguration. It works by reading the target shape signal from the Hub using a Global Receiver, breaking that signal into individual pin requirements (shape type, color, rotation for each quadrant), and then routing raw shapes through processing chains that match those requirements. When the Hub goal changes, the MAM automatically adjusts its outputs to match the new target.
The simplest MAM design uses a "pin-processing" architecture. Each pin on the target shape is processed independently: the MAM produces all needed shape quadrants in parallel, then assembles them at the end. This parallel approach is faster than serial processing and easier to debug when something goes wrong. Most community MAMs use 4 to 8 parallel processing lanes, one for each quadrant of the shape.
Building a MAM is not required to complete shapez 2, but it transforms the late game. Once your MAM is operational, you can walk away from the game and return to find that every Hub goal has been completed automatically. It is the difference between playing an active factory game and having a self-sustaining automation system that runs while you sleep.
MAM Components — Function and Cost
| Component | Function | Cost | Throughput | Placement Tip |
|---|---|---|---|---|
| Global Receiver | Receives the target shape signal from the Hub. Outputs a shape describing what to build. | 50 Iron, 20 Circuit | 1 signal/sec | NE orientation preferred for clean wiring |
| Signal Analyzer | Breaks the target shape into individual pin signals for each quadrant layer. | 30 Iron, 15 Circuit | 1 analysis/sec | Adjacent to Global Receiver |
| Belt Filter | Routes shapes based on pin signal matching. Only passes shapes that match the filter condition. | 20 Iron, 5 Circuit | 1.0 shapes/sec | On every Processing output |
| Painter (Crystal) | Applies crystal paint to shapes. Requires crystal fluid input from Crystalizer. | 40 Iron, 10 Glass | 0.5 shapes/sec | After crystal fluid buffer |
| Crystalizer | Produces crystal fluid from raw crystal deposits. Feeds crystal painters. | 60 Iron, 20 Glass | 1.0 fluid/sec | Near crystal deposit with fluid pipe to painters |
| Stacker | Combines two shape layers. Used to rebuild the target shape from processed components. | 35 Iron, 10 Circuit | 0.5 shapes/sec | Final assembly stage |
| Cutter | Splits shapes into halves for individual processing. First processing stage in most MAMs. | 25 Iron | 1.0 shapes/sec (2 outputs) | Immediately after raw shape input |
| Rotator | Rotates shapes to match target orientation. Controlled by pin signal for rotation state. | 20 Iron, 5 Circuit | 1.0 shapes/sec | Before stackers and after filters |
Step-by-Step MAM Build Order
Build your MAM in layers. Each layer adds one processing stage. Test each layer independently before connecting it to the next. A MAM that fails silently because of a wiring error in layer 3 is extremely difficult to debug after layers 4 through 8 are already built.
| Layer | Objective | Wiring | Common Error |
|---|---|---|---|
| Layer 1 — Shape Cutting | Cut raw shapes into processable halves using Cutter buildings. | No wiring needed at this stage. Pure shape processing. | Forgetting to use both cutter outputs. One output goes to processing, the other to waste or storage. |
| Layer 2 — Color Painting | Apply base colors using Painters fed by Mixers via fluid pipes. | Connect fluid pipes from Mixer to Painter. Add buffer tanks between them. | Painters too far from Mixers. Fluid travels slowly — keep within 10 tiles. |
| Layer 3 — Crystal Application | Apply crystal coating using Crystalizer and crystal Painters. | Crystal fluid pipes from Crystalizer to crystal Painters. Dedicated pipe network. | Insufficient crystal supply. One Crystalizer can supply 2 crystal Painters max. |
| Layer 4 — Stacking & Assembly | Combine processed halves using Stackers. Match target shape orientation. | Pin signals from Signal Analyzer route shapes to correct Stacker inputs. | Wrong Stacker input order. Top layer always comes from the left input. |
Fluid Throughput — Why Valves Matter
Fluid pipes in shapez 2 have a maximum throughput of 1.0 fluid units per second. However, a pipe segment without a valve only delivers a fraction of that — roughly 0.3 to 0.5 fluid per second depending on pipe length. Valves force fluid through at full speed. The community has extensively tested this: you need at least 2 valves per pipe run to achieve full flow, and more for longer pipe runs.
The reason is that fluid in shapez 2 moves in discrete packets. Each valve acts as a pump that pushes one packet per tick. A single valve creates an on-off flow pattern that averages below the pipe maximum. Two or more valves create overlapping push patterns that smooth out the flow and achieve the full 1.0 units per second throughput. This is critical for MAMs because crystal fluid throughput limits how many crystal painters you can run in parallel.
| Pipe Length | Valves Needed | Flow Rate | Notes |
|---|---|---|---|
| 1-5 tiles | 1 valve | Full flow (1.0/sec) | Single valve sufficient for short runs |
| 6-15 tiles | 2 valves | Full flow (1.0/sec) | Add second valve at midpoint |
| 16-30 tiles | 3 valves | Full flow (1.0/sec) | Space valves evenly along pipe |
| 30+ tiles | 4+ valves | Near full flow | Very long pipes may need pumps instead |
Global Receiver Setup
The Global Receiver is the brain of your MAM. It reads the current Hub delivery goal and outputs a shape signal that describes exactly what to build. Place the Global Receiver in the NE orientation (arrow pointing toward the processing area) for the cleanest wire routing. The shape signal exits from the east side of the receiver and should feed directly into a Signal Analyzer for decomposition.
Wire the Global Receiver output to a Signal Analyzer. The Analyzer breaks the target shape into individual pin signals — one for each quadrant and layer. Each pin signal tells one processing lane what shape type, color, and rotation to produce. Route pin signals using wire tunnels under your production lines to keep the surface clear for belt placement.
A common mistake is placing the Global Receiver too far from the processing area. The wire range is generous, but long wire runs are harder to debug when something goes wrong. Keep the Global Receiver within 20 tiles of the first Signal Analyzer. Build your MAM around the Global Receiver location rather than trying to integrate it into an existing factory.
Common MAM Mistakes
- Wrong valve count on fluid pipes. One valve is not enough. You need 2 or more valves per pipe run to achieve full fluid throughput. A MAM with underpowered crystal painting will produce shapes slower than the Hub consumes them, eventually falling behind on delivery goals.
- Insufficient crystal supply. Crystal is the rarest resource in shapez 2. One Crystalizer can supply at most 2 crystal Painters. If your MAM needs more crystal painting capacity, you need more Crystalizers and more crystal deposits connected to your factory.
- Forgotten belt filters. Belt filters are essential for routing shapes to the correct processing lane. Without filters, shapes from one lane contaminate another lane and produce incorrect output shapes. Every processing lane needs a belt filter at its input.
- No overflow management. When the Hub goal changes, the MAM produces a few incorrect shapes before the new signal propagates through the entire system. These incorrect shapes need somewhere to go. Place a Trash building at the end of each processing lane to dispose of wrong shapes.
- Testing the MAM with only one shape type. A MAM that works for square shapes may fail for circle shapes because of belt routing differences. Test your MAM with at least 5 different Hub goals before declaring it complete.
Frequently Asked Questions
Q: How big should my first MAM be?
Your first MAM should target 4 parallel processing lanes (one per quadrant). This is large enough to handle most Hub goals efficiently but small enough to debug when things go wrong. A 4-lane MAM fits in a 40 by 40 tile area. Building an 8-lane or 16-lane MAM as your first attempt is a recipe for frustration — the wire routing alone becomes a nightmare. Scale up once you have a working 4-lane design that you understand fully.
Q: Why is my MAM producing the wrong shapes?
Wrong shapes are almost always caused by incorrect belt filter settings or pin signal wiring errors. Check each belt filter and verify it is set to the correct pin condition. Then trace the wire from the Signal Analyzer to each filter — a single miswired pin can cause an entire processing lane to produce the wrong shape type. Use the wire visualization tool (hotkey: W) to see all connections clearly. Also verify that your Stackers are receiving inputs in the correct order: top layer from the left, bottom layer from the right.
Q: Can I use a Blueprint for my MAM?
Yes, but only for the processing lanes — not the signal routing. The shape processing part of a MAM (cutters, painters, stackers) is highly tileable and perfect for blueprinting. Save each processing lane as a Blueprint module, then copy-paste to expand your MAM. The signal routing (Global Receiver, Signal Analyzer, wire tunnels) is unique to each MAM layout and should be built manually. Never blueprint the entire MAM as a single unit — it becomes too large to place and too difficult to modify.