Becoming a Casting Supplier for Humanoid Robots

If a foundry waits until a humanoid robot manufacturer sends out an RFQ for 100’000 die-castings, the most valuable part of the opportunity is already gone.

The robot industry is currently moving through the stage where hardware architectures are still changing rapidly. The castings are being machined from a block, 3D-printed, and constantly redesigned because manufacturers need flexibility more than low piece prices. At first glance, this makes the market look unattractive for high pressure die-casters. Quantities of 30, 50, or a few hundred castings certainly do not justify an expensive die. However, those prototypes will become tomorrow’s high-volume castings.

The First Generation is Built for Learning

During prototype development, robot companies need to change designs quickly. CNC machining is therefore widely used for structural components because it requires no dedicated tooling and allows the next revision to be produced immediately. 3D printing is used for the same reason, particularly for plastics, jigs and some metal parts.

Cost is secondary at this point because the objective is to get hardware running, test it, and learn from it. A company might produce a few dozen robots and discover during testing that the entire structure needs another revision. Locking such a component into permanent tooling works against the development process. This is very different from the normal environment of a typical automotive program. In humanoid robotics, many decisions are still open.

Prototype, Low Volume, and High Volume need Different Processes

At the prototype stage, machining and additive manufacturing provide maximum flexibility. Once the parts have matured and the company starts producing several hundred or several thousand units, manufacturing shifts toward processes that reduce cost while retaining some freedom. Sand casting and investment casting are the typical technologies that bridge this gap without immediately requiring permanent high-volume tooling.

Only when volumes become large enough does the equation clearly move toward permanent tools and high-pressure die casting. This progression is important for foundries because the customer does not move from one manufacturing technology to another randomly. The part has to survive each step without requiring a complete redesign every time production volume increases.

Design the Machined Part as Tomorrow’s Casting

Humanoid’s approach provides an interesting example. Even when a structural part is still machined from solid material, the design already follows casting DFM guidelines. This reduces the amount of redesign required during industrialization. Instead of optimizing a CNC component perfectly for machining and discovering two years later that the geometry cannot be cast economically, manufacturing considerations are introduced, while the product still has significant design freedom.

For a foundry, this is exactly where the opportunity begins. The sales conversation must therefore not start by asking how many castings the customer needs today. At today’s volumes, the answer may be none. A much more useful question is which machined structural parts the robot company expects to cast once production increases. That opens the door to discussing wall thicknesses, load paths, tolerances, alloy properties, machining allowances, and the future casting process before all the decisions are taken.

The Foundry must enter Before the RFQ

This requires a different kind of business development from the traditional automotive model. Waiting for purchasing to distribute a drawing means waiting until most of the opportunities for differentiation have already disappeared.

The interesting supplier is the one that approaches engineering earlier and helps answer questions such as how a machined part can become a casting, which functions can be integrated, where mass can be removed, which alloy can provide the required properties, and which manufacturing route makes sense at each production volume.

The foundry is therefore selling engineering before it sells casting capacity. That is also why humanoid robotics can be particularly attractive for innovative foundries. The end product is not constrained by decades of established standards. There is considerably more freedom to change geometry and manufacturing methods when it makes sense.

You don’t Need to Produce every Development Stage

A die caster does not necessarily need to start offering CNC machining, investment casting, and sand casting internally to participate in this market. What matters is understanding the complete industrialization path.

A foundry can cooperate with prototype suppliers or low-volume foundries while contributing the DFM knowledge required for the eventual series process. The objective must be to ensure that the casting develops toward a geometry that can ultimately be manufactured economically at scale. This creates a much stronger position than arriving when the high-volume RFQ finally appears. By then, another supplier has already spent two years helping the robot company develop the casting and understands the design decisions behind it.

Follow the Casting, Not the Volume

For me, this is probably the most important lesson from the discussion. Foundries often look at a new market, check today’s annual volumes, and conclude that it is too small to be interesting. Humanoid robotics requires the opposite approach. Today’s volumes matter less than where the manufacturing process is heading.

The component that is CNC machined 30 times today could be sand cast 1,000 times during the next development stage and eventually require a permanent die when the robot enters large-scale production. The transition happens fast. That is why the supplier that supports that development doesn’t have to fight for access once the market becomes attractive. It is already part of the product.

That is the opportunity for foundries. Do not wait until humanoid robots become a high-volume die-casting market. Help create the castings that will make high-volume production possible in the first place.