94 percent of manufacturers plan to raise prices in 2027, and two-thirds tell us pricing is harder to forecast than it was a year ago. Raw materials and labor are part of that story. So are tariffs, which keep moving faster than most procurement teams can re-plan.

Nothing pushes a procurement team to rethink sourcing faster than a cost it can't negotiate away, and custom part-sourcing, which is still less than 1 percent online, needs that push. I saw the same shift at Wayfair during the pandemic and at McKinsey serving many procurement teams, when people who had always bought a sofa in a store went online and never went back. Disruption is doing that for manufacturing.

But most of the tariff debate fixates on where parts get made, even though making parts isn’t the bottleneck. The capacity exists already. About 44 percent of small manufacturers are running below 70 percent capacity.

Where we get stuck is what happens after the parts are made. Look at robotics, where we’ve seen orders spike. A robot isn't built like a car, with everything flowing into one plant for final assembly. Each subassembly pulls from a fragmented supplier base and it must be assembled, kitted and tested in stages before it’s integrated. In the U.S., that multi-step, multi-handoff capability barely exists, so even domestic-made parts move slowly.

Reshoring components without building that layer just moves the bottleneck closer to home.

Policy hasn't caught up. Most of that support goes to materials, parts and components, and very little goes to the assembly and testing that turn them into working machines. I’d like to see us finance small and midsize manufacturers to build subassembly and test capability across the country.

Tariffs will keep changing. What we can control is whether domestic parts become domestic machines.

Sanjeev Singh Sahni, CEO at Xometry