Three EV manufacturing announcements hit my feed this week. Tamil Nadu is the new darling. Pune is bulking up on mobility talent. Tata is pushing upmarket. Every headline celebrates capacity, investment, headcount. Not one mentions quality infrastructure. I heard the same silence at SNOP – standing in a field where a 900-employee plant was supposed to rise, armed with a project timeline and a quality department that existed on a PowerPoint slide.

Ford rehiring 350 engineers because AI quality control could not replace human judgment. That story should embarrass anyone who claims quality can be bolted on, automated, or deferred. It cannot. Stand up an EV plant in an emerging market today, and the first 90 days of quality sequencing will determine whether you survive your first customer audit or surface in a recall notice six months after SOP.

The org chart is the first quality system

Most greenfield plants treat quality as a ramp-up afterthought. They hire a quality manager around week eight, give them a desk near the exit, and expect IATF 16949 compliance to seep out of the concrete. Here is what I did at SNOP.

Week one. Before a single robot is bolted down, you write the quality org chart. Reporting lines. Escalation paths. The first ten standards that govern everything from incoming inspection to nonconformance management. You decide whether quality reports to operations or to the plant manager directly. Choose the former, and you have already lost. I gave quality a seat at the table on day one. Not a chair in the corner.

Staffing sequence matters as much as structure. My first hire was a standards engineer, not an inspector. Someone who could translate IATF, VDA 6.3, and customer-specific requirements into controlled documents before anyone needed them. Inspectors come later, when there is something to inspect. Standards come first. Without them, inspection is opinion with a clipboard.

New geography means untested suppliers

This is the trap nobody talks about. A new EV plant in Tamil Nadu or Pune is not just a new building. It is a new supply chain. Tier-2 and tier-3 suppliers in emerging markets often have thin quality pedigrees. Some hold IATF certificates that look solid on paper and dissolve under a process audit. IATF 16949 does not care that your plant is six months old. The standard has no grace period. Customer auditors will arrive with the same checklist they use in Bavaria or Baden-Württemberg, and they will find what they find.

At SNOP I inherited a supplier base chosen on price and proximity. The first PPAP submissions were a wake-up call. We spent the first 60 days re-qualifying suppliers who had already been approved, because approval without verification is theatre. Run at rate. Capability studies. PFMEA review. The full weight of APQP applied retroactively. It was expensive – consultant hours, delayed launches, strained relationships with purchasing. Not doing it would have been catastrophic.

Quality infrastructure is invisible until it is absent, and by then it is always too late to build it cheaply.

PFMEA culture before volume, not after the first 8D

This is the sequencing decision that separated SNOP from every plant I have audited that became a recall story. We built PFMEA culture before SOP. Not as a documentation exercise to satisfy a registrar. As a living discipline where operators, process engineers, and quality sat in the same room and asked: what could fail, how badly, and what would catch it.

The result: 70% reduction in defect costs. 98% customer satisfaction within the first year of production. Those are not vanity metrics. They are the direct output of sequencing failure-mode thinking ahead of volume. Most plants reverse the order. They run production, discover problems through customer complaints, open 8Ds, and retrofit controls. That path costs ten times more, demoralises the workforce, and trains customers to distrust you.

The Ford AI story is the same lesson at a different scale. You cannot automate judgment you never built. Ford tried to replace engineering expertise with algorithms before establishing whether the quality system underneath could function without humans holding it together. The humans came back. They always come back. The question is whether you hire them on day one or after the recall.

Key takeaways

  • Hire a standards engineer on day one. This person writes the documents everything else depends on. Without controlled standards, every inspector becomes a freelance decision-maker.
  • Make quality report to the plant manager, not operations. If quality reports to the person who owns volume targets, you have built a conflict of interest into your org chart that no audit will forgive.
  • Re-qualify every supplier. Even the ones already approved. A certificate is a starting point. Run at rate and capability studies before the first production lot ships.
  • Run PFMEA workshops before SOP. Not after the first field failure. The cost difference is not incremental – it is the gap between a 70% defect-cost reduction and a customer escalation chain you cannot exit.

The plants announced this quarter in Tamil Nadu, Pune, and wherever the next headline points will succeed or fail on decisions made in the first 90 days that nobody photographs or celebrates. Capacity is visible. Investment is visible. Quality infrastructure is a bet on the boring work nobody sees, made by people who understand that the last plant standing is the one that built its quality function from the foundations up, not from the recall down. I made that bet in an empty field with 900 people counting on it. Morale hit 95%. Not from perks or posters. From giving people a system they could trust. That trust is the real product of those first 90 days. The rest is concrete.