Dark Factory

A dark factory is a factory in which no one is involved in production anymore. Machines process the parts, robotic arms assemble them, driverless transport vehicles bring material from one point to the next. The English name means literally “dark factory”: where no human works, you don’t need light, heating, or break rooms. What is meant is a borderline case of automation, that is, the transfer of work steps to machines. In practice, usually only the actual manufacturing is human-free. Maintenance, programming, and quality control are still handled by humans, often in a normally lit hall next door.

What manufacturers hope to gain from it

Machines don’t know shift changes. A fully automatic line can run around the clock, even on holidays. This significantly increases the quantity a factory produces per year, without needing a new building. At the same time, ongoing costs such as lighting, air conditioning, and safety facilities for personnel decrease.

More important than sheer quantity is often consistency. A robot inserts a component the thousandth time exactly the same way as the first time. For products like computer chips or smartphone housings, fractions of a millimeter determine whether something is scrap. In many cases, human manual work has simply become too imprecise for this.

Behind this, however, there is also an economic-policy reason. In countries with high wages or few young people entering technical professions, workers are lacking. If a factory can manage without personnel, the wage difference compared to other countries hardly matters anymore. That’s why dark factories regularly appear in debates about relocating production back to Europe.

Robots, sensors, and a digital image

For a factory to run without supervision, three things must come together. First, machines that reliably repeat a work step. Second, sensors, that is, measuring devices that constantly report the state of each machine. Third, an overarching software that triggers all steps in the correct sequence.

Artificial intelligence comes into play mainly in two places. Camera systems check every part for scratches or incorrect dimensions and sort out defective pieces. And analysis programs recognize from vibrations or temperatures that a bearing will soon fail. It is then replaced preventively before the line comes to a standstill. This procedure is called predictive maintenance.

Many operators additionally build a digital twin: a virtual model of the complete factory inside the computer. Changes are tried out there before being transferred to the real system. The big catch, however, remains flexibility. A human can pick up a dropped part and improvise; a rigid line cannot. That’s why dark factories so far work best where the same product is made over a long period of time.

Where such factories already exist today

The best-known examples come from Asia. The Chinese manufacturer Xiaomi shows a largely automated smartphone factory in Beijing, and Foxconn has heavily automated several plants. The Japanese robot manufacturer Fanuc has also had robots assembling robots for decades. In Europe, similar approaches can be found in the chip and automotive supply industries.

In the news, you mostly encounter the term in two contexts. Either it’s about the stocks of automation companies and robot manufacturers. Or it’s about the question of how many jobs are being eliminated by such systems. Both debates tend to exaggerate, because a completely human-free factory is, to this day, the exception, not the rule.

A common misconception is that a dark factory manages entirely without personnel. In fact, the jobs merely shift: away from the assembly line, toward maintenance, programming, and data analysis. These positions are rarer but require more training. This is exactly where the real social point of contention lies.