Advanced Manufacturing on the Island: When a Plant Becomes an Engineering Operation

There are two kinds of Puerto Rico manufacturing sites, and from the outside they can look identical.

The first kind executes. Corporate engineering, usually somewhere on the mainland, develops the process. The island site runs it, runs it well, and sends back metrics. When something breaks, a team flies down. When a new product comes, a transfer package arrives.

The second kind develops. The process engineering happens locally. When scale-up problems appear, local engineers solve them. New product introduction is a local capability rather than a visiting one. The site is not a pair of hands. It is where the technical work gets done.

Same buildings, same equipment, sometimes the same company. Radically different economics, and radically different position under the research credit.

Why the second kind keeps winning

The shift from the first model to the second is happening steadily, and it is mostly not driven by tax. It is driven by the fact that the first model is operationally bad.

Remote process ownership creates a lag on every technical decision. The people closest to the equipment are not empowered to change the process, and the people empowered to change it have not seen the equipment in six months. Problems that a local engineer could close in a week take a quarter. Knowledge about why the line behaves the way it does lives in the heads of operators and never makes it into a document, because nobody local has the mandate to formalize it.

Companies eventually notice this, usually after a painful transfer or a yield problem nobody could diagnose remotely. The fix is to put real engineering capability on site.

Once they do, three things tend to happen. Technical issues resolve faster. Transfers get less traumatic. And the site starts generating process improvements that corporate never would have identified, because corporate was not standing there.

The financial side nobody modeled

Here is what usually gets discovered after the fact.

Puerto Rico's incentives framework is Act 60-2019, which consolidated the earlier statutes including Act 73-2008 (Legacy). Under §3030.01, research and development activity carries a credit at two fixed rates: 50 percent on direct R&D investment deployed in Puerto Rico into experimental activities, and 20 percent on indirect R&D investment deployed outside Puerto Rico or into incidental activities.

Read that geographic distinction carefully, because it has a direct operating consequence.

If your process development happens on the mainland and the island only executes, the development spend is sitting in the lower category. If the same development work happens on the island, it is sitting in the higher one. The work did not change. Its location did.

For companies already considering localizing engineering for operational reasons, this tends to tip a borderline decision. For companies that have already localized and never thought about it, it means they may be understating their position substantially.

What counts in a manufacturing environment

Manufacturing companies are the most likely to tell us they do no research, and among the most likely to be wrong.

The qualifying activities in a modern plant usually include process development for new products, scale-up work where the behavior at production scale is genuinely unknown, equipment and automation development where the technical approach has to be engineered rather than purchased, tooling development, material qualification and substitution with real uncertainty, and yield or capability investigations where the root cause is not known at the outset.

Advanced manufacturing adds more. Robotics integration that requires engineering rather than configuration. Additive process development. Inspection and metrology method development. Control strategy and model development where the model has to be built and validated.

What does not count is the daily business of making product. Running a qualified process is production. Preventive maintenance is maintenance. Routine inspection is inspection. A changeover is a changeover. Buying a machine and installing it per the vendor manual is procurement, not development, however large the capital number.

The test, as always, is uncertainty. Did your team know how this would turn out before they started? If yes, it is execution. If no, and they resolved it systematically, you likely have a claim.

The honest difficulty with manufacturing claims

Manufacturing environments have a specific documentation weakness, and it is worth naming.

In a lab, the output of work is a report. In a plant, the output of work is product. The engineering that went into solving a problem often leaves behind only a changed parameter in a recipe, a revised work instruction, and a closed deviation. The intellectual effort evaporates into the process.

That makes substantiation harder than it is for a research organization, and it is why so many manufacturing claims feel thin when examined. The work was real. The record of the work was incidental.

Fixing this does not require a new system. It requires defining development work as projects at the start rather than treating it as part of operations, capturing engineering time against those projects in whatever system you already use, and writing a short technical summary when the project closes that states what was unknown, what was tried, what failed, and what resolved it.

Three paragraphs written contemporaneously are worth more than thirty pages reconstructed two years later. Everybody knows this. Almost nobody does it until they have been through an examination.

What this looks like when it works

The sites that have made this transition well share a few characteristics.

They have a named local process engineering function that is distinct from maintenance and from production support, with its own objectives. They define development projects formally, even small ones. They capture engineering hours in a way that separates development from sustaining. And when a problem gets solved, someone writes down what was learned while it is still fresh.

None of that is exotic. It is ordinary engineering management. The difference is that the companies doing it have a defensible position and the companies not doing it have a story.

Everything here is indicative advisory context. Your actual credit position depends on your decree, your facts, and current administrative practice, and should be confirmed before you plan around it.

The bigger point

The research credit is not a reason to pretend a plant is a lab. It is a reason to be accurate about the fact that modern advanced manufacturing contains a great deal of genuine engineering, and that an enormous amount of that engineering is now happening in Puerto Rico.

Companies that recognize this and organize around it end up with better operations and a stronger position. Companies that keep calling it all "manufacturing support" end up with neither.