Onshoring a Robot Supply Chain: What Actually Cannot Be Sourced Here Yet

The Arithmetic Nobody Does First

The domestic content test that matters for robotics is inherited from the federal acquisition schedule: to qualify as a domestic end product, an item must be manufactured in the United States and the cost of its domestic components must exceed 65% of total component cost for items delivered through 2028, rising to 75% from 2029. That is a cost test, not an assembly test, and the distinction does almost all of the work.

Take a mobile service robot and sort its bill of materials by value rather than by part count. The cheap-and-easy-to-localise group is large but light on cost: chassis weldments, sheet metal, enclosures, castors, bumpers, fasteners, wiring harnesses, some plastics. The expensive group is short and concentrated: the battery pack, the lidar or depth sensing stack, the compute module, motors and gearmotors, motor controllers and drives, and the power conversion electronics.

The uncomfortable result is that you can localise the majority of the parts and still be nowhere near 65% of the cost. A company that ships an imported sensing and compute stack into a United States facility, bolts it into a domestically fabricated frame, and calls the result American-made will not clear the threshold, and on the evidence so far that appears to be the intended effect rather than an oversight.

The public confirmation of this arrived in August, when a company that markets itself as the most American-made robot in its home category stated plainly that it designs and assembles in the United States but does not meet the 65% component benchmark. If a firm built around domestic assembly is short, the test is doing what it was designed to do.

What the Industry Says It Cannot Source

A member survey run by a robotics industry organisation in late August asked fourteen companies, from solo founders to firms above two hundred employees, what is hardest or costliest to obtain domestically. The answers converged on four categories:

• Motors and gearmotors
• Sensors — particularly depth and lidar
• Power supplies and DC-DC converters, with one respondent noting that virtually every affordable unit is made in East Asia
• Robot arms, where a manipulator is part of the platform

Fourteen responses is a soundings exercise rather than industry data, and we would not present it as more than that. But the list matches what we are finding, and it maps almost exactly onto the high-value end of the bill of materials described above. That is the whole problem in one sentence: the parts that are hard to source domestically are the parts that carry the cost, which is why the threshold is hard rather than merely inconvenient.

The same survey found members split close to evenly on whether the listing is beneficial or detrimental, tracking almost exactly with whether a company already manufactures domestically. Worth noting what they asked for: the most frequently requested support was not regulatory relief but a vetted onshoring supplier network, ahead of legal guidance and a longer timeline. That is a request to help firms comply rather than to be excused, and it is the right instinct.

The Uncomfortable Finding About Reshoring

Anyone planning this work should sit with one figure from the 2026 reshoring survey published in September. Across 118 OEM respondents, the share actively reshoring rose to 36% from 29% a year earlier, with 63% planning capital investment in reshoring or domestic expansion. Tariffs were the leading motivation at 65%, geopolitical risk at 60%.

And satisfaction among firms that had already reshored fell to 65%, down from 96% the previous year.

That collapse is the most useful datapoint in the whole survey. It says moving production to the United States is materially harder in execution than in a board paper — supplier depth, qualified labor and landed cost all bite once the plan meets a factory floor. Policy uncertainty was named the primary concern by 57%, which compounds it: you are asked to commit capital against rules that are themselves in motion.

The one encouraging movement in the same data is that the share running a full total-cost-of-ownership analysis rose to 40% from 30%. Firms that model landed cost properly before they move tend to be the ones still satisfied a year later. That is not a coincidence and it is the single most actionable thing in this guide.

Qualifying a Domestic Supplier

Finding a domestic source is the easy half. Qualifying it is the half that consumes the schedule, and for a robot the requirements are stricter than for general industrial equipment because the machine moves autonomously around people.

What a qualification package realistically needs:

• Dimensional and functional conformance against the incumbent part, not against the drawing alone. Drawings are frequently looser than the incumbent supplier’s actual process, and a new supplier building to print can produce a part that is in spec and still behaves differently.
• Performance equivalence under load — for a motor or drive, thermal behaviour and duty cycle at your actual duty, not at datasheet rating.
• Lifecycle evidence. For batteries and drives this is the long pole, and it cannot be compressed by paying more.
• Regulatory and safety documentation that carries into your own compliance file, including anything your notified body or certification path depends on.
• Capacity and second-source posture. A supplier who can make fifty is not necessarily a supplier who can make five hundred, and single-sourcing a critical component to solve a content problem trades one risk for another.
• Country-of-origin documentation you can actually rely on, because your content calculation is only as good as your suppliers’ declarations. Ask for signed declarations with stated maximum age, and keep a record of the gaps.

Budget twelve to eighteen months for a critical component on a platform that must stay in production throughout. Anyone quoting less has either done this before with an unusually cooperative supplier or has not done it.

A Sequencing Plan for a Bill of Materials

A workable order of operations, and the reason each step comes where it does:

1. Build the content calculation before anything else. Sort the bill of materials by component cost, mark each line’s country of origin, and compute where you stand today. Most firms discover they are further from the threshold than they assumed, and it is better to learn that in a spreadsheet.
2. Identify the smallest set of lines that moves you over. This is a knapsack problem, not a sweep. Two or three high-value lines will usually do more than thirty small ones.
3. Separate "hard to source" from "expensive to source." A part available domestically at a 30% premium is a pricing decision you can make this quarter. A part with no domestic source is a development programme. Do not let the two share a line on the plan.
4. Start qualification on the long-lead items first, even if the commercial case is unfinished, because lifecycle testing cannot be accelerated later.
5. Run the total-cost-of-ownership analysis including inventory carrying cost, freight, duty, quality cost and the cost of the qualification programme itself. This is the step 60% of surveyed firms skip and the one that correlates with being satisfied afterwards.
6. Re-run the content calculation quarterly, because supplier mix drifts and the threshold rises in 2029.

We are working through exactly this on our own platforms, and we would rather describe it as difficult than present a tidy timeline we cannot yet defend.

What This Means If You Are Buying Rather Than Building

Most readers of this guide are not manufacturing robots — they are buying them and want to know what the supply side implies for a purchase decision. Four things:

"Made in America" in a sales deck is not the test. Ask for the percentage of component cost that is domestic, measured the way the acquisition rules measure it. A vendor doing this work will have a number. One who has not will change the subject to assembly.

Expect transition-period pricing. A domestic supply chain that does not yet exist is expensive to build, and some of that lands in unit cost. Treat multi-year price protection as a term worth negotiating rather than a courtesy.

Component substitution is coming, and it has technical consequences. When a vendor swaps a motor or a sensor to meet a content target, behaviour can change — runtime, navigation performance in a specific environment, service parts. Ask to be notified of changes to the sensing, drive and power stack, and ask what revalidation you should expect.

If you are a domestic supplier of motors, drives or power conversion and you are not currently selling into robotics, this is an unusually good moment to introduce yourself. The demand is real, the buyers are actively looking, and the industry has publicly said a vetted supplier network is the help it most wants.