Vestas said what nobody in manufacturing wants to hear. The company told its own sector to slow down — not because demand is softening, but because production velocity has outrun the quality systems meant to govern it. That takes nerve. It is also correct, and any quality leader who has stood in a plant watching escapes climb while the volume dashboard gets greener knows exactly why.

The mathematics of scaling instability

Finance departments never internalise this. If you have a process running at 3% defect rate and you double the volume, you do not get twice the good output. You get twice the bad output too — and the cost of that bad output does not scale linearly. It scales geometrically. Each escape carries detection cost, containment cost, customer escalation cost, and reputational damage, and those costs compound with every unit that slips through.

I have watched this pattern across automotive and aerospace plants. A facility pushes volume to hit a quarterly target. The defect rate holds steady as a percentage — 3% — but the absolute number of defective units doubles, the sorting area fills up, the customer complaint queue deepens, and the quality team is firefighting instead of improving. The line was never stable. It was running slow enough that the instability was tolerable. Push it, and the cracks become fault lines.

Vestas is living this. Turbine manufacturing at scale demands precision that leaves no room for a process that has not been characterised, controlled, and verified. When you are building structures rated for twenty-year service life in hostile environments, an escape is not a warranty claim. It is a liability event that can eclipse the unit margin ten times over.

If your quality system cannot pass an audit at current volume, doubling volume does not create capacity. It creates evidence.

Capability before capacity

I built a QA/QC department from the ground up for the SNOP greenfield plant — 900+ employees, new site, new processes, new product lines. The temptation in a greenfield is to push volume immediately. The capex clock is ticking. The board wants to see utilisation numbers. We went the other direction. We stabilised the quality system first: IATF 16949 structure, PFMEA driven into the process design, layered audits, a QRQC discipline that meant problems were contained at the cell, not discovered at the customer.

The result was a 70% reduction in defect costs. Not from going faster. From going correctly. Once the process was characterised and the control plan was real — not decorative — we could scale volume with confidence because the quality system could carry the load. Speed followed capability. It never works the other way around, regardless of how urgently someone needs the numbers.

At Airbus, the work I led on Routing Verification KPIs produced a 97% reduction in internal lead time. That number sounds like a speed story, and in a sense it is. But the mechanism was capability: we built the verification architecture, we instrumented the routing, we made the process transparent. Speed was the output of a system that could finally be trusted. Pushing volume before that system existed would have multiplied the escapes and buried the gains under rework.

Building a quality system that can carry production volume is not glamorous work. It is PFMEA sessions that run long because the team actually argues about failure modes. Control plans that reflect reality, not the version someone typed to satisfy a customer submission. Layered process audits conducted by people willing to find something and write it up rather than nod through a clean report.

Why the boardroom keeps getting this wrong

The scaling trap is not a mystery to anyone who has operated a manufacturing process. It persists because the incentives are misaligned. Quarterly volume targets are visible, measurable, and tied to compensation. Quality system maturity is invisible until it fails, and by then the cost is already incurred and usually assigned to a different budget line.

Look at what just happened at Ford. They bet on AI to solve their quality problems, scaled back the initiative, and rehired 350 engineers. The lesson is not that AI lacks value in quality — I build autonomous multi-agent systems myself, and the capability is real when applied to a stable, instrumented process. The lesson is that you cannot digitise your way out of an uncontrolled process. AI layered on top of instability gives you faster detection of problems you should never have been making in the first place. It is the manufacturing equivalent of installing a smoke detector in a building with no fire doors.

The boardroom metrics reward speed over stability because speed is legible and stability is not. A plant manager who pushes volume and absorbs a 3% defect rate hits the target. A plant manager who holds volume to stabilise the process misses it. Eighteen months later, the first plant is drowning in escalations and the second has a foundation that can out-scale anyone. But by then, the bonus has been paid and the consequence has been reassigned.

Rivian is navigating the same tension right now — growth ambition against the operational discipline required to make that growth survivable. The broader EV sector is learning what Ram Rajappa articulated recently: reliability is not a phase-two concern. It is non-negotiable from day one.

Key takeaways

  • Doubling volume on an uncontrolled process doubles your escapes. Failure cost scales geometrically, not linearly — containment, escalation, and reputational damage compound with each unit.
  • Quality system maturity must precede volume ramps. Stabilise the process, instrument the controls, verify the routing, then scale. Speed built on capability holds. Speed built on hope does not.
  • AI and digital tools amplify whatever state your process is in. They make stable processes faster and unstable processes more expensive. There is no digital shortcut around an uncontrolled process.
  • Board metrics that reward volume without weighting escape costs and audit performance will push organisations into the scaling trap every time. The incentive structure itself is the root cause.

Vestas will take heat for telling the wind industry to slow down. Competitors will frame it as weakness. Analysts will question the timing. Someone will tweet about a lack of ambition. Within eighteen months, the quality foundation they build during that deliberate pause will let them out-scale everyone who kept the pedal down. I have seen this story enough times — in automotive plants, in aerospace supply chains, in greenfield launches — to know how it ends. The companies that stabilise before they scale are the ones still standing when the volume comes. The ones that scale before they stabilise are the ones explaining to the customer why.