The Cheapest Part in Your Factory Is the One That Stops It

Insights from the Scaling Business Summit 2026, Ho Chi Minh City.

Most conversations about factory automation start with the robots. The AGVs, the AI vision systems, the autonomous production cells. Christoph Dirr, General Manager at Würth Industry Vietnam, started with the fastener.

It wasn't a detour. It was the point. After 25 years supplying industrial manufacturers across 60 countries, Christoph had a clear-eyed view of what actually stops a factory. Not the big systems. Not the expensive machines. The commodity parts nobody thinks about until they're gone.

1. The Smallest Parts Cause the Biggest Shutdowns

Most factories spend serious time optimizing major procurement decisions. They negotiate hard on large-ticket items. They benchmark suppliers. They build rigorous approval processes for significant spend. And then they largely ignore the 4,000 low-value parts that keep the line running every day.

Christoph Dirr called this out directly. C-parts - fasteners, PPE, MRO consumables, industrial chemicals carry almost no individual unit value. But the complexity of buying them, keeping them in stock, and getting them to the right worker at the right moment is enormous.

"The challenge of a factory nowadays is to have 4,000 parts that have a relatively small value in your whole value chain, but come with a very high complexity of buying, keeping on stock, and making available to the worker at any given point of time."

The financial logic is straightforward once stated. A line-down event caused by a missing bolt that costs a few cents generates opportunity costs that dwarf any savings from switching suppliers or running lean on inventory. Dirr put it in terms that are hard to forget: "C parts are like the blood that pumps through the ecosystem of a factory. As soon as you don't have it anymore, you have a heart attack."

For operations leaders, this reframes the entire commodity procurement conversation. The goal isn't to find the cheapest fastener. It's to guarantee the fastener is there.

Lesson 1: Stop optimizing C-parts on unit price. Optimize on availability. A line-down costs more than a year of supplier savings.

2. Real-Time Visibility Is Not Foresight

The current state of the art in factory supply management is RFID-enabled kanban - a system Toyota invented decades ago, now upgraded with sensors and data transmission. Two bins of every material. One empties, you replenish. RFID made that replenishment signal faster: instead of waiting for a bin to hit zero, factories now get hourly consumption data from floor-level sensors.

This feels like intelligence. It isn't. Christoph was precise about the distinction: "When the bin is empty, you will order a new bin. It's a reaction. It's not something that you anticipate." Every IoT system deployed at scale today, RFID readers, smart vending machines, sensor-tagged storage produces data about what already happened. The supply chain responds to the past.

That's still valuable. Hourly visibility is better than daily. Daily is better than the paper card a worker fills out at shift end. Transparency reduces waste, improves stock management, and makes limited resources go further. But factories that mistake current-state visibility for predictive capability are building on a false assumption.

The next step, what Christoph described as the real frontier is purchasing that turns proactive. Demand forecasting built from production plans, not consumption history. Supply positioned before it's needed, not after it's used.

Lesson 2: Real-time data tells you what happened. It takes production-plan integration to tell you what's coming. Know which one you actually have.

3. AI Needs a Data Floor Before It Can Have a Ceiling

When the Q&A turned to AI, Christoph gave an answer that cut through most of the hype in the room. The bottleneck for AI in manufacturing isn't algorithms. It isn't compute. It's the absence of structured data from the physical operations where decisions actually get made.

"First you need data and the capacity to store it and to access it very fast. You can have the best hardware if you do not know what to feed it."

The practical version of this looks like: a factory whose workers write material usage on paper cards produces no usable data. A goods-to-man picking system without sensor coverage has no feedback loop. Smart demand-management software, the kind that Blue Yonder or process-mining companies like Celonis are building can only work if there are data points to process. Without them, the algorithms have nothing to train on and nothing to optimize.

Christoph described the first version of proactive supply as modest but real. Cross-referencing production plans against bill-of-materials data across thousands of similar manufacturers to forecast what a factory will need before they order it. "We can ask the customer: how many cranes do you plan to build this year? And from that we can calculate how much they actually need." It sounds obvious. It requires years of structured data to execute.

Lesson 3: Audit whether your physical operations are generating usable data before evaluating any AI layer. No data infrastructure means no intelligence, just expensive software running on noise.

4. Vietnam Does Not Have to Follow the Same Path

One of the session's sharper observations was about how Vietnam's supply chain development might unfold. Christoph flagged a pattern from China: for decades, Chinese banks tried to move consumers from cash to credit cards to e-banking. Nobody made the transition. Then WeChat arrived, and the entire population moved to mobile payments in a few years, skipping the credit card era entirely.

The implication for Vietnam's industrial supply chain: it does not need to follow the same step-by-step evolution that advanced European or North American economies went through. It could skip intermediate stages and move directly to a platform economy model with invisible inventory. “What will happen in Vietnam does not need to follow this pattern.”

Vietnam's supply chain has a specific structural challenge: most C-parts are imported. Approximately $80 million in fasteners enter Vietnam every month. Local manufacturing of these parts is shallow and not particularly versatile. The supply chain is more complex and fragile than in China, where dense local supplier networks provide flexibility. That complexity makes automation and data management not just useful but necessary. “Vietnam really needs a lot of this and especially data.”

Lesson 4: Vietnam's supply chain constraints are real, but so is its leapfrog potential. The companies that design for where the supply chain is going, not where it has been will build a structural advantage.

The CEO Execution Playbook: What to Do Tomorrow

1. 1. Map your C-part failure risk. Identify the five cheapest parts in your bill of materials that would cause a line-down if missing. Then check whether their replenishment is visible, automatic, or still managed manually. That gap is your first automation priority.
2. 2. Classify your visibility honestly. Identify which of your current IoT or inventory systems are reactive (triggered by depletion) versus proactive (triggered by production plans). If the answer is "all reactive," that's your gap, not your achievement.
3. 3. Run a data audit before any AI conversation. Before evaluating demand-management software or forecasting tools, confirm that your physical operations are generating structured, timestamped consumption data. If workers are still using paper, start there.
4. 4. Build planning discipline around your supplier constraints. Map the concentration of your supply base. For any category where fewer than five suppliers hold most of the volume, model what a disruption costs and build minimum lead times accordingly.

→ Watch the full session on YouTube