Tooling as strategy: how to scale industrial innovation without burning millions
Small is beautiful (again): startups in the age of steel and semiconductors
Industrial startups: the emergence of a new class of venture
The very phrase industrial startup feels, at first glance, contradictory. Or at the very least, an oxymoron.
In the late 1990s and early 2000s, startups embodied the ideals of a post-industrial economy: light, agile, asset-light. The story always started in a garage. Their value lay in code, data, and networks — not in machinery, factories, or physical goods. The startup archetype scaled through servers, not supply chains.
It grew by shipping features, not shipping containers.
Industry, on the other hand, invoked heavy capex, multi-decade depreciation cycles, tooling, complex labor relations, and regulatory oversight. It was a domain of incumbents — steel mills, chemical plants, automakers — not of two founders with a pitch deck.
And yet, the boundaries have shifted. Today’s most pressing challenges — decarbonization, resource efficiency, resilient manufacturing — are physical by nature. They require innovation in materials, energy, and production processes. And so, startups are increasingly venturing where VCs once feared to tread: into the world of atoms.
But this raises a series of fundamental questions: what does it mean to build a startup in an industrial context? And how do we finance such ventures without falling into the traps that doomed many of their predecessors?
Put differently, how do you turn a contradiction in theory into a success in practice?
The costly lessons of the first industrial startup wave
The good news is that the past decade has given us plenty to think about. Many early industrial startups tried to replicate the hypergrowth playbook of Silicon Valley — raising massive rounds, building giga-scale infrastructure, and promising to reinvent entire value chains overnight.
But this strategy, seductive on pitch decks, proved structurally flawed in practice.
In the past decade, we’ve seen industrial startups attempt to scale through sheer capital accumulation. Their logic: raise hundreds of millions, build massive infrastructure, and gain market share through first-mover advantage.
You can hardly blame them: it was the startup playbook of the 2010s, shaped by the rise of major digital platforms. The logic of these giants was very much in the zeitgeist, and it naturally rubbed off on the early industrial ventures just finding their footing.
The results have been sobering, to say the least:
Ÿnsect raised over $625 million to build vertical insect protein farms[1]. Despite technical breakthroughs, it failed to achieve economic viability at scale. Tooling, facility commissioning, and bio-process ramp-up proved far costlier and slower than anticipated. By 2024, it sought bankruptcy protection[2].
Northvolt, hailed as Europe’s battery champion, has raised more than $13 billion across debt and equity[3]. Yet its giga-scale plants have suffered cost overruns, supply chain disruptions, and delays that have set back production timelines by years[4]. Tooling up a gigafactory for next-gen cells is neither trivial nor cheap: each assembly line alone represents hundreds of millions in custom equipment.
Innovafeed secured over $500 million to industrialize insect-based ingredients[5]. But despite scientific progress, its economics remain challenged by high capex, complex tooling for bioreactors, and the difficulty of entering commodity markets at profitable margins.
These examples share a common flaw: scale-first thinking without validated economics. By locking themselves into expensive, inflexible tooling and mega-factories from the outset, they sacrificed agility and invited capital dependency.
Tooling: the invisible weight of industrial scale
At the heart of these challenges lies an often-overlooked concept: tooling.
In an industrial context, tooling encompasses the custom molds, dies, jigs, fixtures, and production equipment required to make parts or products at scale. It is what translates a CAD model into physical reality — repeatedly, reliably, and at unit costs that make business sense.
For startups, tooling represents a structural tension:
Soft tooling (e.g., 3D-printed molds, temporary fixtures) allows for prototyping and iteration at low cost — but is unsuitable for high volumes or tight tolerances.
Hard tooling (e.g., steel injection molds, precision dies) enables mass production — but demands major upfront investment (often €50,000–€500,000 per tool[6]) and locks in design choices.
This is where many industrial startups falter. They commit too early to expensive tooling for unvalidated products. Once millions have been sunk into custom equipment, pivoting becomes prohibitively expensive.
Moreover, tooling often implies supply chain entanglements: long lead times (12–24 weeks for complex molds[7]), reliance on overseas suppliers, and exposure to geopolitical risk.
A smarter path focuses on modular, flexible tooling strategies, delaying hard tooling investments until unit economics are proven at smaller scale.
The new industrial playbook: capital-efficient, modular, and resilient
The next generation of industrial startups understands these dynamics — and builds accordingly. Their hallmarks:
Leveraging existing infrastructure
They retrofit into existing plants, supply chains, or production lines — avoiding unnecessary capex.Iterative, modular scaling
They prove unit economics at pilot scale before committing to hard tooling or large-scale builds. They design production systems that can grow organically, module by module, rather than requiring giga-scale bets.Precision on unit economics
They obsess over the breakeven point of each process step, knowing that in industry, scale amplifies inefficiency as much as it does efficiency.
Case studies: Spark Cleantech, Everdye, Diamfab
We see these principles embodied in the industrial startups we back at Asterion:
Spark Cleantech – Produces low-cost hydrogen without multi-billion-dollar plants. The model can be characterized as a clean-energy-as-a-service model. By harnessing waste industrial heat, Spark achieves 4x efficiency gains over conventional methods[8], all without major tooling investment in electrolysis stacks or new infrastructure. Rather than demanding costly overhauls, Spark’s modules plug into existing infrastructure—no retrofits, no disruption, and from a customer’s perspective, dramatically lower entry costs.
Everdye – Aims to decarbonize textile dyeing, responsible for ~20% of global water pollution[9]. Its proprietary dyes integrate into existing dye houses, avoiding both capex-heavy plant redesign and the need for new tooling. Energy savings? Up to 10x per unit produced[10].
Diamfab – Develops diamond-based semiconductors for high-power electronics. Where competitors chase billion-dollar fabs, Diamfab demonstrates that diamond substrates can be grown with capital-light reactors, eliminating the need for giga-scale semiconductor tooling.
These ventures succeed because they respect industrial realities: tooling matters, capex discipline matters, and scale should follow market validation, and not precede it.
Venture capital designed for industrial logic
At Asterion, we apply the same logic to our own operations. Instead of raising massive blind funds, we deploy capital via SPVs, building alignment around each deal, minimizing dilution, and avoiding the financial equivalent of unnecessary tooling: excess capital that invites waste.
We focus on:
Industrial businesses where capex is right-sized
Teams that understand tooling and production constraints
Models where scale is earned, not assumed
Just as modern industrial startups reject over-building, we reject over-funding. The goal is the same: resilient, capital-efficient growth.
Final thought: smart capital beats brute force
Whether in AI — as DeepSeek recently showed by matching GPT-4 with a fraction of the compute[11] — or in industry, the future belongs to those who can do more with less.
Tooling, capex, infrastructure: these are not evils to be avoided. But they must be approached with discipline, timing, and precision.
The next generation of industrial champions will not be those who raise billions first. They will be those who master when and how to invest in the physical machinery of scale.
Small is beautiful. And when it comes to industrial startups, small (at first) is often what survives over the long term.
[1] Crunchbase – Ynsect
[2]: Les Echos – Ynsect in Receivership
[3]: Northvolt Investor Relations
[4]: Reuters – Northvolt factory delays
[5]: Innovafeed on Crunchbase
[6]: Protolabs – Tooling Cost Estimates
[7]: Fictiv – Tooling Lead Times
[8]: Internal Asterion data from Spark pilot projects
[9]: UN Environment Programme – Textile Pollution
[10]: Everdye technical documentation (2024 white paper)
[11]: DeepSeek’s Release Notes
Three questions for… Philippe Berlan, CEO, Everdye
Philippe Berlan spent his career leading iconic French brands — from Petit Bateau to La Redoute — before joining Everdye in 2024. As CEO, he now leads the industrial scale-up of a deeptech textile company. A case study in doing more with less — and exactly the kind of trajectory this piece set out to explore.
R&D investments are one of the first pitfalls on the path of an industrial startup. How did you approach this issue at Ever Dye?
From the very beginning, we made a strong choice: not to confuse R&D with industrialization. The goal wasn’t to demonstrate our technology under ideal conditions, but to quickly validate its viability in real-world settings.
So we worked with a tight-knit, highly specialized R&D team and iterated as closely as possible to industrial constraints: colors, fibers, processes.
Rather than investing in our own production lines, we used standard machines, under real conditions, with field partners. The result: controlled costs, proven reproducibility, and a very clear roadmap. We prefer to move forward brick by brick, each one validated, rather than trying to cover everything at once.
Which means you had to validate the solution commercially at a very early stage… What convinced your first industrial clients to commit?
The mistake would be to think that environmental impact alone is enough. What made the difference is that we didn’t ask them to change a thing. No capex, no training, no dedicated line.
Our technology fits into existing equipment — and more importantly, it improves their process. Less energy, less water, fewer chemicals, with the same or even better result. That immediate compatibility lifted the final barriers.
And once the Adore Me capsule was successfully launched in the US, it validated the model — both commercially and operationally.
You're now entering the industrialization phase — arguably an even trickier pitfall. How are you tackling it?
We won’t be building an Ever Dye factory anytime soon… maybe not ever! We chose a capex-free, outsourced (CMO) model to maintain agility and scalability. Scale-up happens module by module: one pigment, one fiber, one process. Every time we validate a new building block, we scale it with industrial partners.
At the same time, we're optimizing our production costs by playing with pigment concentration, material yield, and tank size.
We know exactly when our technology becomes competitive without a “green premium.” We’ll reach that point in 2026, with margins continuing to grow as we scale. The goal isn’t to move fast. It’s to move right — with solid fundamentals and a long-term industrial vision.
The future isn’t just coded — it’s built (TED 2025)
At TED2025, industry and manufacturing took center stage. Lauren Dunford argued that innovation isn’t just happening in code — it’s happening on factory floors. From AI to everyday goods, everything still needs to be built. To succeed, we need to reinvent manufacturing as a platform for progress, not a symbol of decline.
Techno‑industrialist: YC’s call for an industrial renaissance (Y Combinator)
Y Combinator’s latest “Techno-Industrialist” RFS calls for startups to bring modern software to physical industries: energy, materials, biotech. The goal? Smarter, faster, greener systems at scale — and a revival of industrial strength with a startup mindset. A signal Europe can’t afford to ignore.
An intelligent path to reindustrialization (MIT Tech Review)
MIT’s Initiative for New Manufacturing outlines a smarter model: integrating AI, data, and sensing into production itself. Think low-carbon materials, predictive maintenance, human-machine collaboration. More than just tech — a new industrial operating system, with public–private alignment at its core.