If i understood correctly what FT said that the main problem is to get enough electromagnetic excitation power in the THz-range. Their solution is to create very sharp, high dU/dt (or dI/dt), pulses and wish enough harmonic components in the high frequency area. It is easy to understand the problem, because the highest electric frequencies in normal life are those found in WiFi or inside a microwave oven, a few GHz. So 1000x more is needed. I checked a little bit in the internet what is available at the moment. The fastest components seem to be capable of handling a few hundreds of GHz but THz is still far-away.
To succeed they need to invent something quite exotic to overcome the problem. Fortunately there is quite a lot of interest in this frequency area especially in sensor development so maybe suitable components are soon available there ?
Sparking would give them those frequencies much easier. I am sure they are aware but timing is probably also key. It may be complex to time sparks with a nanosecond accuracy.