Logo

If the steel survives the code

Old freezers, emergency buttons, and airplanes: hardware lasts for decades, but IT becomes obsolete in a matter of months.

Published on November 3, 2026

coding

Team IO+ selects and features the most important news stories on innovation and technology, carefully curated by our editors.

A supermarket lost 100,000 euros worth of food because the cooling systems suddenly set the temperature to 50 degrees. It wasn’t due to a broken compressor; there was no refrigerant leak, and the power grid remained stable. Mechanically, the systems were in top condition; with some regular maintenance, they could easily have lasted another fifteen years. But the software couldn’t make it that long.

The culprit was hidden in a small gray box on the wall: the control unit. Malicious actors had gained access through outdated systems from suppliers such as Danfoss and Dixell. The controllers were connected directly to the public internet, secured with factory default passwords like “12345” and software for which no security patches had been released in years. A single command from outside was enough to turn the freezers into ovens.

This incident is not an isolated one. It is the most visible manifestation of a creeping crisis in technology: the inevitable clash between the lifespan of hardware and that of software.

Two worlds, two speeds

Modern industry operates at two speeds that are fundamentally at odds with each other. On the one hand, there is the world of mechanics and operational technology (OT). An industrial cooling system, a transformer substation, an airplane, or a car is designed with durability in mind. Maintenance follows the laws of classical physics: you replace the bearings on time, change the oil, and check the gaskets. As long as you do that, a machine in 2026 will continue to function exactly as it did when it was delivered in 1996.

On the other hand, there is the fast-paced dynamic of IT. Software does not wear out due to friction or heat, but because of the relentless changes in its environment. Cryptographic standards are cracked, communication protocols become obsolete, APIs are phased out, and network infrastructures are simply dismantled. Unlike mechanical wear and tear, this is usually not a gradual process. Everything works until it stops working. And this wear and tear also doesn’t mean replacing parts with exactly the same parts—just new ones.

A machine that sits idle in a shed for thirty years and is then started up again can usually be made to work with a little care. A computer program that remains untouched for five years often can no longer be started up because external dependencies no longer exist. Even if it does work, there’s a good chance it barely functions and is more of a security vulnerability than a useful application.

As long as those two domains remained separate, there was little to worry about. But over the past twenty years, the drive toward digitization has led us to cram every piece of copper, steel, and concrete with microchips, modems, and internet interfaces. We’ve connected devices with a life expectancy of half a century to digital umbilical cords that start to rot after just five years.

The falling 2G dominoes

Look beyond the supermarket freezers, and the cracks in the system are becoming visible everywhere. The most acute illustration of this is currently unfolding in our mobile networks. Telecom providers are rapidly shutting down the old 2G network to free up spectrum for 5G. Technologically, this is a perfectly logical step, but for the physical world, it’s a slow-motion disaster.

In the Netherlands, grid operators like Liander are forced to prematurely remove hundreds of thousands of perfectly functioning “smart” energy meters from meter cabinets. There’s nothing wrong with the meters; they measure electricity consumption just as accurately as they did on the day they were installed. But because their internal GPRS modems will soon be unable to find a cell tower, they’ll be reduced to useless scrap overnight.

The same drama is unfolding with the tens of thousands of personal alarm devices that allow vulnerable seniors to call for help with the push of a button. Does the button work? Mechanically, yes. Does the call go through? Soon it won’t anymore.

Even the automotive industry faces a potential debacle. The European Commission has already warned of gaps in the so-called eCall system—the built-in emergency button that automatically alerts emergency services in the event of a serious traffic accident. In millions of relatively new cars, that module will soon be unable to establish a connection, simply because automakers continued to install cheap 2G and 3G modems well past 2018. The airbags will still work for decades, but the car will be deaf and mute.

Floppy disks in the cockpit

The aviation industry demonstrates just how bizarre this mismatch can become when digital components are expected to function for decades. A Boeing 747-400 is a technical masterpiece of aluminum and jet engines that can effortlessly remain operational for forty years. But to provide the onboard computers with essential navigation and airport data every month, maintenance technicians still have to climb into the cockpit with a bag full of 3.5-inch floppy disks.

Those floppy disks haven’t been mass-produced anywhere for years. Airlines have to scour the country for old stockpiles on auction sites to keep state-of-the-art aircraft safely in the air. Installing a modern USB reader or a wireless module sounds trivial, but it faces sky-high costs for recertification by regulatory agencies. And so the hardware remains trapped in the digital Stone Age.

In the consumer sector, the industry solves this in a more pragmatic—but more cynical—way: by pulling the plug. When Signify decided to pull the plug on the first-generation Philips Hue Bridge, the device immediately lost its online functionality. The hardware—a circuit board and a plastic casing—could have functioned for decades more. Instead, an entire generation of home automation ended up in the landfill because the device no longer met modern cloud standards.

The modular taboo

How are we going to solve this problem? Slowing down IT won’t work; throwing away perfect hardware after five years is economically and ecologically irresponsible.

The answer lies in the way we approach engineering. Anyone who buys a car expects to be able to replace a worn-out tire or a dead battery without sending the entire chassis to the scrapyard. But when it comes to digital components, manufacturers have virtually abandoned the concept of modularity.

Chips, modems, and operating systems are soldered directly onto the motherboard or tucked away deep inside sealed enclosures. A slot for a network card—which would allow you to replace an outdated 2G modem with a 5G or Wi-Fi module in a snap—is almost always missing in consumer electronics.

With industrial devices, however, this is often possible. For example, you can install a modern Danfoss control panel on a 2005 refrigeration system without changing a single bolt on the freezer unit itself. But because franchisees found that update too expensive and the control box was still working, the digital backdoor remained wide open.

Within industrial automation, there’s also a prevailing fear:“if it’s working, don’t touch it”. Upgrades are avoided out of fear that a running factory will come to a standstill, causing outdated devices to rot away digitally behind the firewall for years—at least, if you’re lucky. The affected freezers were simply left wide open on the internet.

In consumer electronics, manufacturers don’t even want to make that distinction. There, the modem is soldered directly onto the circuit board, meaning the entire device must be scrapped as soon as the network protocol changes.

This is rarely a matter of technical inability; it is a deliberate economic choice. A closed, monolithic circuit board costs a few dimes less during the production process than a modular design with standardized connectors. Moreover, it fits seamlessly with the business model of planned obsolescence.

The price of the illusion

The hack of the Danfoss cooling units shows that we have reached the limits of this model [6]. The romantic notion that you can install an industrial machine and expect it to function for twenty years just as it did on day one is over.

It’s time for regulators and buyers to face this reality. Just as the European Union is now requiring manufacturers to provide replaceable batteries and universal USB-C chargers, there must be a strict standard for modular electronics in capital goods.

If a device has a design life of twenty years, the digital control layer simply cannot be permanently fused to the mechanics.

If we fail to do so, we’ll remain trapped in a vicious cycle in which we send perfectly healthy machines to the scrap heap just to replace an outdated chip—or leave the doors wide open to anyone who wants to corrupt our food supply chain with a simple script. Hardware may be built to last forever, but as long as IT remains mortal, we must learn to separate the two.