The manufacturing revolution

Raw materials in, robots out.

We're building the software and hardware that turn raw materials into complete, mission-ready robots. No assembly lines. No fragile supply chains. No compromises. Just exactly the machine the mission needs, and as many as the world needs.

Hardware · Raw matter to robotsSoftware · Software Defined MatterAustin, TX

01 · The problem

Machines are still built like it's 1926.

Every robot moves on actuators: the powered joints that combine a motor, gears, and electronics to drive an arm, a leg, a wing. Strip one open, even on the most advanced robot in the world, and you'll find a century-old recipe: hand-wound copper coils, laminated steel cores, precision-ground bearings, rare-earth magnets. Dozens of parts from dozens of suppliers, assembled by hand.

Every joint is a small supply chain. Every interface between parts is a future failure point. And the magnets at the heart of nearly every actuator on Earth come overwhelmingly from a single strategic competitor.

The robotics revolution is being built on a manufacturing foundation that can't support it.

50+

separately sourced parts in a single robot actuator

90%

of rare-earth processing is Chinese-controlled

100 yrs

since the basic electric-motor recipe last changed

Path one · Catch up

Onshore yesterday's actuator

Replicate the proven designs, import the machinery, rebuild the magnet and winding supply chain domestically. Necessary work, but it ends in parity at best, on a cost curve our competitor set.

Path two · Leapfrog

Our path

Change what an actuator is

Delete the assembly, the interfaces, and the magnet dependency. Don't race the old supply chain. Make it unnecessary, and let robotics grow as fast as it can be designed.

03 · The software

Software Defined Matter

You can't grow a machine you can't describe.

Leapfrogging the actuator starts in software. Software Defined Matter is our engineering framework for designing machines in code: the entire path from specification to working machine is designed and executed by software, and the output is a physical, multi-material object.

No hand-drawn CAD. No hand-authored meshes. Geometry, materials, objectives, and constraints are all code, and all differentiable, so the whole design loop is one graph that can be optimized end to end. CAD, evolved.

01 · Define

Write the intent, not the drawing.

A Computational Engineering Model captures what the machine must do: specifications, parametric geometry, materials, objectives, and constraints, all in code.

02 · Standardize

One file every tool can read.

The model compiles to a .sdm file: a portable description of shapes, materials, objectives, constraints, and how parts couple. Parameters survive the trip, so the next tool can still move them.

03 · Evolve

Let the physics shape the part.

Gradients flow from a physics objective back through the shape itself. The design is optimized, validated, assembled, and exported as machine-ready geometry for the process that will make it.

build_part.py→ hollow_cylinder_with_hinge.sdm

# geometry is a signed distance field, not a mesh
outer  = sdf_primitive("capped_cylinder", h=30.0, r=ref("outer_radius"))
inner  = sdf_primitive("capped_cylinder", h=30.0, r=ref("inner_radius"))
shell  = sdf_op("subtract", [outer, inner])
infill = sdf_op("intersect", [shell, sdf_primitive("gyroid", period=8.0)])
hinge  = sdf_primitive("notch_hinge", notch_radius=ref("notch_radius"))

part = Part(
  name="hollow_cylinder_with_hinge",
  materials=[
    MaterialRegion(name="Cu",  sdf_tree=hinge),
    MaterialRegion(name="PLA", sdf_tree=infill),
  ],
)
save(part, "hollow_cylinder_with_hinge.sdm")

Render of a multi-material hollow cylinder: a red copper top disk with a hinge over a blue PLA shell with gyroid lattice infill — One part, two materials, no mesh authoring. A copper hinge over a PLA shell with a gyroid infill, defined entirely by the code beside it.

Isometric render of a bar that is solid red at one end and solid blue at the other, with the two materials dispersed into each other across the middle — Graded, not joined. One solid piece of two materials, with composition following Fick's second law across the join instead of a bonded interface.

Board of signed distance field primitives: gyroid and other TPMS lattices, compliant mechanisms such as notch hinges, leaf springs and bellows, and lofts and sweeps — A vocabulary for matter. Lattices, compliant mechanisms, lofts, and sweeps, each an exact signed distance field that composes with the rest.

Software · Software Defined Matter

Intent in.

A specification becomes a machine-ready design: geometry, materials, and physics optimized together, so every machine is designed for its mission instead of adapted from a catalog.

Hardware · Growing the machine

Raw matter in.

The process and machines that turn raw materials into that design: complete, multi-material robots, with no assembly line and no fragile supply chain between them and the mission.

05 · The vision

This is robotics' silicon-chip moment.

The silicon chip didn't just secure computing, it unleashed it. When circuits stopped being hand-assembled, computers stopped being rare. The same is about to happen to robots.

When a complete machine can be designed and manufactured from raw materials, robotics stops being rationed. Every farm, clinic, shipyard, and launch pad gets exactly the machine its mission needs: a Cambrian explosion of robots, designed in software, grown from raw matter, built on a foundation no one can pull out from under us.

Build exactly the machine the mission needs.

Working on robotics, defense, or advanced manufacturing? We'd like to hear what you need built.

hey@emergentmatter.com