- The world's smartest machines still depend on one of its dumbest networks
- The AI revolution stands atop a 35-year-old cellular network
- Innovation moves fast. European regulation still takes the train
Trains are really cool, aren’t they? Except when they’re late or break down. Then they suck.
Germany’s national railway recently moved to the "suck" end of the customer service ratings when its entire network ground to a shuddering halt for two hours on June 23.
Not cool, Deutsche Bahn. Not cool at all.
But wait, it gets worse (or better, if you aren’t German and are enjoying the schadenfreude). It turns out the root cause wasn't a sophisticated cyberattack or sabotage. It was a routine maintenance mistake involving a technology many people assumed had disappeared with the flip phone: 2G radio.
Somewhere, a German engineer — let's call him Klaus, and pretend he’s smoking a giant meerschaum pipe — is quietly pondering how his 2026 railway was brought down by a system belonging to the same technological era as the Nokia 3210 and the Spice Girls.
Germany's railway failure was not simply a rail problem. It was a reminder that much of the world's critical infrastructure still depends on technologies that are decades past their intended lifespans.
While Big Tech massively overhypes an AI future and commits trillions to data-center build-outs, the invisible machinery that keeps modern industry functioning remains quietly tethered to second-generation mobile networks that debuted in 1991, delivering the raw, hair-raising power of 9.6 kbit/s. Woof! Slow down, Hans!
Buy a new car in Europe and its legally mandated eCall emergency system will probably still communicate over 2G. Millions of electricity and gas meters do the same. So do ATM monitoring systems, elevator emergency phones, industrial sensors and countless remote monitoring devices.
The reason is political, not technical. Europe mandated 2G eCall before LTE-M and NB-IoT — the cellular standards specifically designed for IoT devices — were ready. By the time those superior technologies arrived, the legislation had already locked manufacturers into yesterday's solution. And this is Europe. Rules are rules.
It illustrates one of Europe's most persistent industrial weaknesses: innovation moves faster than regulation, yet regulation often proves far harder to modernize than the technology itself.
Taking the breaks off
The United States has increasingly erred in the opposite direction, adapting existing regulations to accommodate emerging technologies rather than creating new ones. Neither extreme is ideal. One risks freezing obsolete infrastructure in place; the other risks allowing critical systems to evolve faster than the regulatory frameworks designed to protect them.
The nuclear industry provides a cautionary example. Rather than develop AI-specific rules, U.S. regulators have opted to treat artificial intelligence as another software tool subject to existing regulations, allowing it to be deployed in operational nuclear power stations without a dedicated AI regulatory framework.
America is currently doing this whole "make up your own rules" thing a lot. Whether it involves starting wars or trying to fix World Cup games, it seems to think it can just do what it wants without consequence. But its latest war has been a series of embarrassing faceplants, and it lost the football game. What does that say about its infrastructure choices?
Choose your poison
So which is better? Germany's "vee vill make zee rules" bureaucratic authoritarianism, or America's "screw it, just send it!" approach to regulation?
It depends on your perspective. Personally, I'd rather my train wasn't late. But I'd also rather not live in an irradiated wasteland because Big Tech believes it has a God-given right to use probabilistic-guesswork LLM chatbots to manage sensor data at the local nuclear power station.
The fact is that some governments regulate too much. Others regulate too little. Almost none have found the regulatory Goldilocks zone — where the rules evolve fast enough to keep pace with technology without stifling innovation or compromising safety.
2G or not 2G?
But back to 2G, because this is where the story goes global.
As an industry, we spent years celebrating the retirement of 3G, assuming the hard work was done. It wasn't.
2G is the honey badger of the telecom industry: hidden, tough as old boots and astonishingly difficult to eliminate. While everyone celebrated the retirement of 3G, the real legacy network — 2G — quietly remained embedded throughout modern society in well over a billion network endpoints.
The obvious question is why operators don't simply switch it off. It's because replacing 2G is a lot more than a software upgrade; it is a civil engineering project.
Millions of smart meters must be replaced. Millions of vehicles require upgrades. Remote industrial sensors often have to be physically located, excavated or replaced after years in the field.
Telecom operators therefore face an uncomfortable dilemma. They want to reclaim valuable 2G spectrum for more capable and profitable 5G services. But switching it off too quickly risks disabling critical infrastructure and disconnecting vulnerable people from essential emergency systems.
Ironically, 2G survives not simply because replacing it is expensive, but because it remains remarkably good at the job it was designed to do. Its low-frequency signals travel farther and penetrate buildings better than many higher-frequency cellular bands, while simple monitoring devices can run for a decade on a single battery. Modern LTE-M and NB-IoT can match or even exceed that performance — but only after millions of legacy devices have been replaced.
Germany's railway disruption demonstrates why this situation cannot continue indefinitely.
The hardware supporting these legacy systems is reaching the end of its life. Spare parts are becoming scarce. The engineers who understand them are retiring. Europe's railway sector does not expect to complete its migration to the 5G-based Future Railway Mobile Communication System until well into the 2030s.
That should concern far more than railway operators.
Twenty-nine countries have already retired their 2G networks. Germany is not among them. For a country that built its reputation on engineering excellence and industrial efficiency (not to mention making the trains run on time), that is an uncomfortable statistic.
Germany is now trying to accelerate infrastructure modernization. It is a welcome change of direction. But governments and telecom operators everywhere must stop treating the retirement of 2G as a routine network upgrade. It is a national infrastructure program requiring coordinated regulation, sustained investment and long-term planning.
Because for all the excitement surrounding artificial intelligence, the machinery that keeps modern civilization functioning still depends on a cellular standard introduced more than three decades ago.
The world isn't running on AI. It's still running on 2G.
And until we replace it, we remain only one maintenance error away from the next systemic failure.
Stephen M. Saunders MBE is a communications analyst and USPTO-registered inventor examining how digital infrastructure — 5G, cloud, and AI — is reshaping industry, power and society, as well as underpinning the emerging, ubiquitous global digital economy. As anchor of FNTV and a longtime industry insider, he focuses less on growth narratives and more on execution, risk and how hyperscale technology is distorting markets, governance and society at scale.
Opinion pieces from industry experts, analysts or our editorial staff do not represent the opinions of Fierce Network.