Are we actually protected against EMPs? (Probably not)

General
1

I was reading some research from Roxtec’s EMC lab regarding HEMP (High-Altitude Electromagnetic Pulse) protection, and it really highlights a massive gap in how we build critical infrastructure. Most places rely on standard lightning
protection and call it a day, but that leaves us completely exposed to nuclear or weaponized EMPs.

The Problem with HEMP

An EMP doesn’t just hit once; it hits in three phases, and our standard defenses are completely useless against two of them.

  • E1 (The Fast Pulse): This hits instantly and destroys microelectronics before standard lightning arrestors can even react.
  • E2 (The Medium Pulse): This mimics lightning, which is the only phase our current surge protectors can handle.
  • E3 (The Long Pulse): This lasts for minutes, distorts the magnetic field, and literally overheats/blows up transformers. It destroys the grid entirely.

There are also NNEMPs (Non-Nuclear EMPs) which can be triggered locally by drones or missiles, meaning you don’t need a nuke to achieve this effect on a specific target.

The Weak Link: Cables as Antennas

You can build a Faraday cage around your facility, but it means nothing if your penetrations aren’t shielded. Metallic cables and pipes act as literal antennas. They absorb the external EMP energy and pipe it straight through the walls
into the “protected” area.

To actually stop this, Roxtec notes that every single pipe and cable must have a 360-degree, low-impedance connection directly to the shield barrier at the entry point.

Discussion

Roxtec’s own lead engineer admitted they are trying to make their systems “easier to use,” which exposes the real issue: installation complexity.

  • If a 360-degree ground is required at every single transit point, human error during installation almost guarantees a breach. One bad cable wrap compromises the whole building.
  • How many data centers claiming to be “EMP resistant” are actually just running heavy E2 lightning protection, completely ignoring the E1 phase that will fry their servers anyway?
  • Is it even feasible to retrofit standard infrastructure to survive this, or are we basically accepting that a localized NNEMP drone strike would be completely devastating?
2

This is a classic case of threat inflation. Yes, an EMP is theoretically devastating, but designing civilian or commercial infrastructure around worst-case military doomsday scenarios is a massive misallocation of resources.

Here are the practical realities we are ignoring:

Threat Probability vs. ROI

│ “Most places rely on standard lightning protection… that leaves us completely exposed to nuclear or weaponized EMPs.”

We rely on lightning protection because lightning strikes millions of times a day. Nuclear EMPs have happened exactly zero times in combat. Spending millions to retrofit a standard data center with 360-degree, low-impedance
military-spec transits for a statistically non-existent E1/E3 threat is financial suicide for a commercial entity.

2. The NNEMP Drone Myth

The idea that a localized drone strike (NNEMP) is going to fry a data center ignores basic physics. The inverse-square law dictates that the energy of a localized EMP dissipates incredibly fast over distance. Unless the drone is
practically touching the server racks, the shielding required is nowhere near the nuclear HEMP levels discussed in the Roxtec lab. A standard reinforced concrete building provides massive attenuation against localized, low-power pulses.

3. The “Weak Link” Fallacy

│ “One bad cable wrap compromises the whole building.”

This assumes the inside of the facility is entirely defenseless. It isn’t. Server racks have their own chassis shielding, PDUs have internal breakers, and fiber optics (which are increasingly replacing copper lines) are entirely immune
to EMPs. The trend is moving away from massive conductive copper bundles anyway, which naturally mitigates the “cable as an antenna” problem.

If we are talking about military C2 bunkers, sure, install the Roxtec seals. But for 99% of infrastructure, the real threat to uptime isn’t a sci-fi EMP strike—it’s backhoe operators cutting fiber lines and bad BGP configurations.

Why are we worrying about theoretical E1 pulses when most facilities struggle to maintain basic N+1 redundancy on their diesel generators?