A few weeks ago, I sat across from the COO of a European luxury brand. Thousands of employees, factories on multiple continents, a name you’d recognize. I asked him a question I ask every industrial executive I meet: what’s your white-collar to blue-collar ratio, excluding R&D?
His answer: 1.2 to 1.
Let that land. For every person making the product, there are 1.2 people planning it, scheduling it, reporting on it, coordinating it, approving it, tracking it, and emailing about it. This is a manufacturing company. The people who don’t manufacture anything outnumber the people who do.
He’s not an outlier. He’s the norm. And this is the single biggest problem no one in manufacturing wants to talk about.
As always : I’m renan, I run OSS Ventures, the venture builder for the industrial world. We create software for factories based in Paris and Boston. The views below do not represent financial advice and are an honest reflection on our day to day work.
The Reshoring Illusion
The reshoring narrative has momentum. Tariffs are going up. Supply chains got burned during COVID. Governments are subsidizing domestic production. The political will is there. The strategic logic is there. So why isn’t it happening faster?
The standard answer is labor cost. Western workers cost too much, the argument goes, so factories will always drift toward lower-cost geographies. This is the answer you hear at conferences, in policy papers, and from consultants who haven’t run a factory.
It’s also wrong. Or rather, it’s incomplete in a way that makes it useless.
The labor cost problem is real, but it’s not where you think it is. A modern automated factory doesn’t need an army of blue-collar workers. Between robotics, automation, and improved process design, the shop floor headcount in a reshored factory is manageable. Not cheap, but manageable. The economics can work.
What kills the reshoring math is everything above the shop floor. It’s the planners, the schedulers, the quality coordinators, the compliance officers, the procurement specialists, the production controllers, the ERP administrators, the reporting analysts. The white-collar infrastructure that wraps around every factory like a bureaucratic exoskeleton.
You can automate a welding station. Up until recently, you couldn’t easily automate the six people who coordinate, schedule, report on, and argue about that welding station. And when those six people earn Western salaries instead of emerging-market salaries, the reshoring business case collapses — not because of the workers on the line, but because of the people in the offices above it.
This is counterintuitive. It defies the mental image most people have of a factory. But the data is consistent across every industrial I work with. The white-collar ratio is the hidden tax that makes Western manufacturing uncompetitive, and almost nobody is framing the problem correctly.
The Leverage Shift
There is a way out. But it requires a mindset change that most industrial organizations are not ready for.
The current model treats white-collar roles as linear. You have 14 factories, you need 14 planning teams. You open a new production line, you hire a new coordinator. Headcount scales with complexity, roughly one-to-one, and that’s accepted as a law of physics.
It’s not a law of physics. It’s a systems failure.
Which is weird, because factory directors I talk to can cite a simple ratio immediatly : number of machines per blue collar. CAPEX per blue collar. What about the rest ?
The organizations that are starting to figure this out are making a fundamental shift. Instead of many white-collar workers with low leverage, they are moving toward few white-collar workers with enormous leverage — paid significantly more, equipped with systems that multiply their reach, and expected to operate at a completely different altitude.
I’m seeing the early signs of this shift in the companies I work with. A US-based industrial runs 14 factories with 5 planners. Not 14 planning teams, not even 14 planners. Five. They do this because they invested in systems — real systems, not a PowerPoint about digital transformation — that give each planner visibility and control across multiple sites simultaneously.
A European knitwear manufacturer is exploring how to move from 50 R&D engineers to 5. Not by cutting corners on product quality, but by giving a small team of exceptional engineers tools that let them do in hours what previously took weeks of manual iteration. Generative design, simulation, AI-assisted pattern development — not as buzzwords in a strategy deck, but as daily working tools.
These are not hypothetical scenarios. These are real conversations happening right now, with real companies, about real headcount decisions. The ratio is shifting from 50:1 effort to 5:1 effort, and the people who remain are neither cheaper nor less skilled. They’re more expensive, significantly more capable, and dramatically more productive.
The mental model that needs to die is this: “We need a person for every function.” The mental model that needs to replace it: “We need a system for every function, and a person to govern ten systems.”
Why It’s Hard
If the answer is so obvious — fewer people, better systems, more leverage — why isn’t everyone doing it?
Two reasons, and neither is technical.
The first is that the systems are genuinely hard to build. Enterprise software in manufacturing is a graveyard of failed implementations. ERPs that took three years to deploy and never worked properly. MES systems that the shop floor ignores. BI dashboards that nobody trusts. The scar tissue is deep, and industrial leaders have learned through painful experience that “just deploy the system” is a fantasy. Our recent experiences of successes show that AI must be the tool to deploy : data ingestion (largely solved now), know-how formalization (genAI 10x this). Implementation without AI embedded in the activities is bound to fail.
This is a real obstacle, but it’s dissolving faster than most people realize. Done well, the current generation of AI-native tools is fundamentally different from the enterprise software of the past decade. They’re lighter to deploy, faster to show value, and designed to augment the user rather than replace their judgment. Of course, that is if a true product person thougt the deployment process, not someone creating a product for the sole purpose of justifying consultant fees (did you know, most big consultancies take a backfee on the software they deploy ? talk about adverse incentives). The technological barrier is lower than it’s ever been if done right. It’s not zero, but it’s no longer the binding constraint. And AI in this process finishes the solving.
The second reason is harder. It’s people.
Telling an organization that it needs to go from 50 white-collar workers to 10 is not a systems problem. It’s an identity crisis. Those 50 people have careers, families, institutional knowledge, and political influence. The middle managers who oversee them have built their authority on headcount. The HR function is designed to hire and retain, not to compress and upskill. The entire organizational culture is built around the assumption that more people equals more capacity.
Reversing that assumption is a leadership challenge of the highest order. It requires executives who can simultaneously make the economic case (“we cannot compete at this ratio”), the human case (“the people who stay will have better careers”), and the technical case (“the systems exist to make this work”). Most leadership teams can make even one of these arguments. Very few can make all three at once with enough conviction to actually move the organization.
Last, the work those people do is categorically different from the previous work. So it’s massive upskilling also.
This is why the transition is happening slowly, from the edges. The companies making the shift are either founder-led (where the CEO can make unpopular decisions without a committee), in genuine economic distress (where the status quo is no longer an option), or structurally advantaged (where they built lean from day one and never accumulated the overhead and have a lot of courage and can think in first principles).
What the Future Factory Looks Like
The factory of 2035 doesn’t have fewer blue-collar workers than today’s factory. Automation has been chipping away at shop floor headcount for decades, and that will continue at its own pace.
The real transformation is upstairs.
Here’s a prediction : the factory of 2035 has a white-collar to blue-collar ratio of 0.3 to 1, not 1.2 to 1. It has a planning team of 3 people running what used to require 30. It has a quality function of 2 people with AI systems that catch more defects than the team of 15 they replaced. It has no reporting analysts, because the data flows directly from the machines to the dashboards to the decisions without a human translating spreadsheets in between. Oh, and those factories can operate from anywhere because the fundamental equation of volume to number of humans has been changed.
The people who remain in these roles are not today’s white-collar workers doing less. They’re a different profile entirely: systems thinkers, comfortable with AI tools, able to manage by exception rather than by process, and paid accordingly. A planner overseeing 14 factories through AI-augmented systems is not an administrator. That person is an operator of a complex system, and they should be compensated like one. Those salaries should rival banking or bigtech.
This is not a future that arrives by buying software on the shelf, chosen by a committee of people who don’t want to change their jobs. It arrives by rethinking what a manufacturing organization is for. The question is not “how do we make our current people more productive?” The question is “how many people do we actually need if we build the right systems?”
That second question is uncomfortable. It challenges headcount as a proxy for capability. It threatens established hierarchies. It forces honest conversations about which roles create value and which roles exist because the systems weren’t good enough to eliminate them.
But it is in our opinion the question that separates the manufacturers who will thrive in the next decade from the ones who will slowly become uncompetitive, buried under the weight of an overhead structure designed for a world that no longer exists.
The factory’s real unlock was never on the shop floor. It’s in the offices above it.
You guess the floor.