No, parent scales are not always major.
First, what is a parent scale, actually?
I think the usage of the term "parent (major) scale" is mainly due to how we have historically been taught the diatonic scales - I'd wager that most people start by learning the Ionian (major) scale first, next it's usually Aeolian (minor), then the rest.
To remember how to construct the rest of the diatonic scales (in 12TET), we learn that these are derived from a specific order of half-steps and their cyclic permutations; meaning that if we start from the major scale: [2, 2, 1, 2, 2, 2, 1]
, we can derive the dorian scale: [2, 1, 2, 2, 2, 1, 2]
by moving the initial whole step to the end of the list.
This commonly taught way of constructing/deriving the scales means that we usually have to think of a parent scale first, which is — out of habit-- commonly the major scale.
What's the alternative?
We can enumerate the 12TET notes from 0
to 11
and arrive at the definition of a pitch class set:

Here, the "initial" whole step that gets "rotated" around is highlighted in green, and the flattened/sharpened notes (from the perspective of the major scale) are given in bold.
According to the works of Forte and Rahn, we now examine these pitch class sets, and stipulate that one of them is the "prime form", while the others (the modes) can be derived purely by rotations. The prime (or "normal") form is chosen to be the one which is most compact, in this case the Locrian mode.
In this context, therefore, we might call the Locrian mode the "parent scale" of the diatonic scales, which evidently is not a major scale. Note however, that due to the way we're taught, most musicians instinctively rather think of modes of the "major scale".
Further reading
Here, I'd strongly suggest to check out Ian Ring's great website, which lists all 2^11=2048
scales with a lot of analysis and clickable midi examples. Most importantly, he found that by representing a scale in binary form major scale: 101010110101
(picture this as if each note on a 12-tone keyboard being on or off, reading from right-to-left!)
we can directly derive a decimal representation major scale: 101010110101 == 2741
and if we do this for all the modes, we can find the "most compact set" is the one with the lowest decimal number (the locrian scale).
So what does this tell me about any non-major parent scales?
Using Ian's scale finder, it's extremely easy to find the modes of any scale; take for example the Ultralocrian scale with its modes: Locrian Natural 6, Major Augmented, Lydian Diminished, Phrygian Dominant, Aeolian Harmonic, Harmonic minor scale, all of which are clearly not the major (i.e. Ionian) scale.