"This is a very real threat," he said. "EMP is generated when a small nuclear weapon, 40 to 60 kilotons or about three times the size of a Hiroshima bomb, is detonated 200 miles out in space above the United States. It sets up an electrostatic discharge which cascades to the Earth's surface, feeds into the millions of miles of wires which become antennas, feeds this into the power grid, overloads the grid and blows it out."
This is not quite accurate but let's run with the claim for a minute.
This was first recognized as an issue during a high-altitude nuclear test. But one must be very careful to generalize for a number of reasons, not the least of which is that said test was a thermonuclear device with a yield of 1.4 megatons.
In other words it wasn't a small or tactical weapon -- it was a hydrogen bomb and a pretty good-sized one. Modern tactical nuclear weapons are typically somewhere between 1/10th and 1/100th that size, simply because it was recognized that the inverse square law is real and setting off really big ones doesn't scale the damage on a linear basis.
As noted in the paper cited there are two "waves" in an EMP; one is directly caused by the detonation and the second is essentially the coupling of same into the magnetic field. The latter is basically the same thing as what occurs during a serious geomagnetic storm caused by CMEs from the sun -- sort of, except that a nuclear detonation is a point-source of energy and a CME is not.
We do not have modern-era empirical evidence of exactly how serious of a threat this is, simply because atmospheric (and those in both space and underwater) tests were banned by international treaty (and have been observed quite well) in the years since 1963 -- well before the modern electronic era.
But do not kid yourself that EMI damage potential has been ignored in the years since. It most-certainly has not. Your car, for example runs around with extremely high voltages in the ignition system and without pretty-decent shielding and bypass the electronics it would be immediately fried. Indeed original CMOS ICs, which were not protected internally very well at all, were known to be blown up when handled if you looked at them wrong, and were shipped in carbon-bearing conductive foam for this reason.
The second element that arises is that there are "natural chokes" that are present in the electrical grid and elsewhere. A "choke" is simply something that inhibits the passage of higher frequencies while allowing lower frequencies to pass (or even coupling them.) A transformer is one such example. The risk of a high-frequency transient passing through one is thus that it is high enough voltage to violate the insulation -- that is, to pass via arcing rather than through it. That risk is real but each time that occurs a decent amount of the energy is shunted since some of it will go to ground rather the other side.
Again we can't empirically test any of this -- but what we do know is that when that test was performed while Hawaii experienced disruption their electrical system was not destroyed and further, it didn't nail anything over here on the mainland or elsewhere. The effect, of course, was unintentional as well because we didn't know it would happen.
One thing we should keep in mind however, is that de-centralization is good and we've destroyed a lot of that. Smaller municipal-size generators, many of them coal-fired, have been turned off and most of those have had the wrecking ball taken to them. That adds risk because prior to that you could, for example, pick up a metro area after an event very rapidly even if you can't reconnect to the grid at-large until other areas were able to be restarted. Some plants after a trip, particularly large fission nuclear reactors, often cannot restart immediately (with fission plants this happens due to Xenon poisoning and depending on the age of their fuel load they can require a day or two before the plant can re-establish criticality) and those plants also cannot "black start" at all by design -- that is, they cannot come up on their own without grid power because their emergency generators are only sufficient for emergency shutdown cooling water flows and not normal startup.
Of similar concern is the fact that when it comes to larger transformers we simply do not have material spare capacity that is not connected and in-use. Unless you're willing to, on a coordinated basis, disconnect and ground those units on detection of a possible launch (and you better not miss one) then unlike a solar event which we do "see" well in advance of it getting here having a material percentage of those units being destroyed by internal arcing is certainly possible. That many of these aren't made in the US and there is no US supply is even more-crazy, but that's the world we live in today -- and we should definitely correct that problem.
Certainly a coordinated, multi-warhead event like this would be considered an all-out declaration of nuclear war with all that entails. But a single event, while of concern, is not going to end life in America -- or destroy all our infrastructure. It just won't, and while an all-on assault of this sort would be extremely bad so is blowing up a hundred large cities and both are, for all intents and purposes, the same sort of act-of-war when you get down to it.
Yes, we should take this seriously, but let's be realistic about the circumstances under which that actually occurs and what it means rather than considering it an "isolated" sort of event.
It wouldn't be and we wouldn't treat it as if it is.