The technology transfer is complete. The process knowledge isn't.

Three aerospace expansion announcements landed in the same week. KAI signed an MOU with Embraer on aircraft structures. Diehl Aviation opened a new component facility in Craiova. India's precision manufacturing story got another breathless write-up about how aerospace contracts turned the country into a machining and assembly powerhouse. Each came with a signed document, a press release, a rendering of a clean new building. None came with a plan for the thing that determines whether the parts conform: the process knowledge that lives in one engineer's fifteen years of adjusting feed rates by feel, never written down because he was too busy keeping the line running.

I see this from the inside. My remit at Airbus covers manufacturing engineering technical authority across North America, and a regular part of my week is watching what actually moves between sites and what stays behind in someone's notebook. The drawings travel. The control plans travel. The PFMEA — if you're fortunate — gets updated before it travels. But the knowledge that prevents the first defect stays in the building where it was earned, usually in the head of someone who wasn't invited to the transfer meeting.

What the contract transfers and what the shop floor needs

A technology transfer agreement is a legal instrument. It specifies what data package changes hands, what licensing fees apply, what IP boundaries exist. Thorough in the way contracts are thorough — it covers everything that can be litigated and almost nothing that makes a part conform on a Tuesday at 2 AM when the heat-treatment furnace is running three degrees hot and the operator on shift three has been qualified for six weeks.

On paper, the receiving site gets 2D drawings, 3D models, material specifications, process flow diagrams, control plans, work instructions, tooling designs, inspection standards, and a validation report from the source site. Sometimes a digital twin. Sometimes a full A3 from the last improvement cycle.

What the receiving site actually needs to achieve first-pass yield on day one of serial is every unwritten adjustment the source operator makes without thinking. The dwell time that isn't in the work instruction because it was never measured — the operator just knows the structural adhesive needs forty more seconds when humidity exceeds sixty percent. The visual inspection boundary between "acceptable witness mark" and "rework" that the source inspector applies based on twelve years of seeing parts that passed and parts that didn't. The specific way the CNC program needs modification after a tool change because the tooling supplier changed the insert geometry six months ago and nobody updated the nominal program.

I built the quality function for a 900-person greenfield plant at SNOP. Equipment, standards, training systems — all achievable with enough capital and a clear IATF 16949 roadmap. The hard part was embedding process knowledge that prevents the first defect. Clean facilities, calibrated equipment, certified operators, and still the first six months produced escapes that traced back to things nobody had thought to write down. Not negligence. Tacit knowledge is invisible until it's absent.

The year-one escape window

Every technology transfer has a quality valley. Roughly twelve months. That's where the escapes live.

The pattern repeats across sites, programs, companies. First quarter: validation, expatriates from the source plant on site, parts produced under intense scrutiny. Metrics look strong. Everyone is in the room. Then the expatriates leave and serial production ramps. The control plan runs as written. The first escapes appear — not catastrophic failures but subtle dimensional drift, surface defects near the specification edge, bond-line inconsistencies the source site never saw because the source operator did something that was never in the work instruction. By month seven the receiving site is in reactive mode, drowning in 8D reports, trying to reverse-engineer process knowledge that should have arrived as capability, not as documents.

I drove a fifty percent reduction in EASA audit findings in one cycle at Airbus. Most findings in a transferred process trace back to the same root cause: the receiving site was given a photograph of the process instead of the process itself. A control plan describes what to check. It does not describe what to feel, what to listen for, what to adjust when the process drifts two sigma before the SPC chart catches it.

The cost of this gap is not abstract. Each escape in aerospace generates a quarantine, an investigation, a containment action, and sometimes a customer escalation that consumes engineering hours at rates that would make a consulting firm blush. A single problem resolution report from a tier-one aerospace customer can burn through six figures in direct cost before the root cause is identified.

Transferring capability, not paper

The fix is not more documentation. More documentation is what organisations reach for when they don't know what else to do. The fix is structured capability transfer — a deliberate, engineered process that treats tacit knowledge as a deliverable, not a residual.

The MOU makes the trade press. The first production escape makes the quality alert. Build for the second.

In practice: source operators embedded at the receiving site for the full year-one window, not three months of validation support. Process failure mode analyses conducted jointly — source and receiving engineers walking the line step by step, with the receiving engineer asking the questions the documentation can't answer. QRQC cadence established before the first escape, not after. PFMEA reviewed line by line with the people who run the process, not the people who wrote it in a conference room. And a living register of "things we do that aren't in the work instruction," treated as a controlled document and maintained as part of the process specification.

None of this is expensive relative to a year of escapes. All of it requires something most technology transfer plans don't budget for: time from the people who actually know the process, spent at the receiving site, for longer than anyone wants to pay for.

Aerospace manufacturing is decentralising faster than its process knowledge can follow. New facilities in Romania, South Korea, India — necessary, welcome, building capacity the industry genuinely needs. But capacity is not capability, and a signed agreement is not a conforming part. The next time you read about a technology transfer, ask the question the press release won't answer: what is the plan for the knowledge that has never been written down? That knowledge is what makes the part. Right now, in most transfers, it isn't in the scope.