According to the developer, the detectors' discrimination works well throughout its depth range.
This is true for every detector in totally benign soil. This isn't true for every detector in soil that isn't entirely benign.
As the depth of the target increases, the magnitude of the target's impedance vector reduces (or the magnitude of the targets response). Once the magnitude of the target's response approaches the magnitude of the ground signal, the detector will lose its ability to determine the ferrous / non-ferrous properties of the target. In other words, for all detectors, a sufficiently deep target / weak response begins to be distorted by the surrounding ground. All detectors lose the ability to discriminate before they lose the ability to detect the presence of a target. This is basic physics / linear algebra in action. This is true for Minelab, XP, Nokta and yes Nexus and literally every other VLF in existence. This is a general limitation imposed by the physics of the magnetic field being transmitted into the ground and has nothing to do with SF/SMF nor Digital / Analogue.
So Georgi isn't technically lying, but he isn't being totally honest either about how the detector will behave in most ground conditions. Note that even in Calabash's very mild soil, the ground balance is incredibly finicky.
Is the Nexus "pinpoint based"? Pinpoint to me, means motion, no discrimination, and short blips on small targets. The Nexus has the opposite of all of that.
I am not sure if this is a typo, or if we are assigning different meaning, but pinpoint is literally a non-motion mode. E.g. you can hold the coil stationary above a target and it will continue to make a tone.
This is what the Nexus does, but with the caveat of slowly detuning / attenuating over time. The XP Deus 2's pinpoint mode can behave in the same way when you enable the detuning, albeit it detunes faster.
For example, an SMF detector can be disced to only hit on copper and silver, which is a massive amount filtering, yet little if any depth loss occurs when doing so.
This is audio filtering, meaning it happens in the audio processing layer. A digital detector can take this signal, store a perfect copy in memory, and then runs filters on it without distorting the original copy stored in memory. An analogue detector takes this signals and runs it through a chain of several hardware filters in sequence, each hardware filter distorts the signal slightly by introducing noise along the way. This means for analogue, more filters means less depth. For whatever reason, this concept that filters distort the original signal is a bit of a vestige from analogue days. This is why people incorrectly think, for example, that turning on/off the horeshoe button on a Minelab detector affects its depth.
Rather, I think it's mainly a result of an abnormally low recovery speed.
The Nexus very slowly detunes. This is the 'low recovery speed' that Georgi is referring to. This technology isn't anything new, and has been around for longer than I have been alive.
Recovery speed can be thought of as transient amplification. E.g. when a signals rate of change exceeds some value (positive derivate), we amplify the signal, but likewise, when it begins to decrease (negative derivative), we can suppress that signal faster in preparation for a different nearby target to begin to increase the rate of change back to a positive derivative. E.g. a low recovery speed suppresses a decreasing signal much slower than usual (this is the slow detuning of the Nexus).
The Nexus isn't doing anything novel, or groundbreaking. All detectors designed to go very deep, e.g. XP Extreme, 2 Box detectors, have exactly the same 'very low recovery speed' mode.