Two Years for a Wear Part

Why hardware development takes so long — and why that isn't a flaw in the process

There's a question I've been hearing for three years now, in every possible tone of voice. Friendly, impatient, sometimes pitying. It goes: So when are you finally going to sell something?

It's a fair question. It's just an incomplete one.

Because it assumes that selling is the hard part and everything before it is a kind of preparation. In software, that may well be true. In hardware, it's the other way around.

A component that looks like nothing

At the front of our boring robot sits an inductive sensor. In a photograph you can barely make it out a piece of technology inside a drill head, unremarkable, a few centimetres across.

Its job: detecting unexploded ordnance. Metal in the ground, before the machine runs into it.

On the cost sheet, that's one line. In my life, it's two years.

Why the sensor has to be there at all: anyone drilling in Germany is often drilling into ground with a history. Former industrial sites where pipes, foundation remnants and storage tanks have disappeared over decades without anyone keeping a reliable record. Converted military land. Contaminated layers. And across large parts of the country, ordnance that has been sitting in the soil for eighty years and appears on no map.

For a boring robot that goes underground without line of sight, that isn't a side condition. It's the core requirement. Humanitarian demining would be the logical technical extension it simply isn't the market we're building the product for. Our market is trenchless utility installation. The sensor is what makes automated work there permissible in the first place.

The same sensor, second job

There's another reason the sensor sits exactly where it sits: it detects the target pit and shuts the machine down.

That sounds like a detail. It is in fact the difference between remote-controlled equipment and an autonomous system.

With conventional small-bore technology, someone stands at the target pit for the entire duration of the bore, watching for the moment the head breaks through. One person, one whole bore, waiting for something that happens eventually. That's expensive time and depending on the pit, not a particularly pleasant place to spend it.

BOB has its line, follows it autonomously, recognises the target and stops. With nobody standing there.

That's where the commercial value actually comes from: not from the machine being able to drill, but from it finishing the job on its own.

The physics nobody has on their radar

The sensor doesn't sit just anywhere. It sits inside a heated drill head.

The first reaction is almost always the same: so you melt your way through the ground?

No. The heating isn't there to melt. It's there to dry.

The chain behind it is simple once you've seen it:

Wet soil sticks. Sticky soil clogs the auger flight. A clogged helix has no advance. And a machine with no advance is a very expensive ground anchor.

So: dry first, then convey.

That sounds like one sentence. In reality it's a conflict of objectives that runs for months. The heating needs energy and installation space. The sensor needs a clean electromagnetic environment and a heating element right next to it is roughly the opposite of that. Both sit at the point of the machine that takes the highest load, is the hardest to reach, and is most likely to break.

Every solution to one problem made the other one worse. For two years.

That's not a complaint. It's simply an honest description of what happens when you develop a component that has to do several things at once that get in each other's way.

And then comes the punchline

This drill head is a replaceable part.

Not something you casually unscrew, granted. But it is a wear part at the tip of an entire machine. It gets replaced eventually, the way a bit, a chisel or a tooth on an excavator bucket gets replaced.

Two years of development for something that is swapped out as a matter of routine.

Anyone coming from software finds this absurd. Anyone who has built hardware nods wearily. Because that is exactly where the difference lies that the question about selling overlooks:

You can patch software. You can't patch a tool.

Whatever ends up in an injection mould stays in it. A deploy costs nothing; a new mould costs five figures. A bugfix reaches every user overnight; a design error is sitting in every machine already in the field and has to be repaired there one at a time, with travel, spare parts and downtime at the customer's site.

That's why it takes time. Not because we're slow. Because every detail has to be right once, for the next ten years.

What's happening under our feet right now

There's a second reason this time isn't wasted time. And it occupies me more at the moment than any delivery date.

We're building technology that dries the ground while outside, the ground is turning to dust on its own.

The droughts of recent years aren't a weather topic for us, they're an engineering topic. Dry soil behaves differently from wet soil. Different friction. Different borehole stability. Different behaviour under advance. Sand that previously passed as cohesive soil now trickles back in.

Anyone designing ground technology today is designing it for conditions that may no longer exist by the time the product reaches the market. That isn't a theoretical risk. It's the reason our drill head has a heating circuit that can be regulated, rather than one that is simply on or off.

Nobody sees decisions like that from the outside. They appear in no product brochure. But they determine whether a machine still works in five years, or whether it was built for a world that no longer exists.

Why we still have to talk about selling

So there's no misunderstanding: selling matters. Without revenue there is no second generation of machines, no series production, no company. Anyone who romanticises development and looks down on sales builds very beautiful prototypes and goes very reliably bankrupt.

But selling is one part of the overall process. Not the process.

The other part is the years in which nobody sees anything. In which suppliers drop out, components go end-of-life, a prototype falls apart in testing and you drive home with an insight that sets you back three months.

Those years appear on no balance sheet. They appear in the product.

If you'd like to see it for yourself

At GaLaBau 2026 in Nuremberg, BOB will be on the stand and turning. Not a rendering, not a video a machine you can walk up to and put your hands on.

15 to 18 September 2026, Nuremberg Exhibition Centre, Hall 3C, Stand 3C-415b.

We have a contingent of free day tickets.

Link: https://n8n.srv855592.hstgr.cloud/form/2494d376-c5f8-4697-8a05-4731f2d360e7

Do bring your specific application ground conditions, pipe diameters, access situation. We'd much rather talk about that than about data sheets.

See you in Nuremberg.

Nicolai Stern Managing Director, Stern Technic GmbH

Stern Technic develops autonomous small-bore technology for trenchless utility installation. No trenching, no guesswork. More at sterntechnic.com.