In most companies they are in five or six places at once, and none of those places is the system of record. Finished videos tend to sit in the shared drive, in whatever asset manager the brand team bought, and on the platform where they were published. The footage those videos were cut from sits somewhere else again: an editor's drive, an agency's storage, a hard disk in a drawer, or inside an edit project that only opens on the machine it was built on. Which of those you are looking in decides what you can realistically find.
The finished cut and the footage are two different questions
A finished video is a single file with a name, so it survives reasonably well. Someone can usually produce a copy of it, even if the copy is a compressed export rather than the master. Source footage behaves differently. It is large, it was never meant to be shared, and it gets pruned whenever a drive fills up. If the thing you want is a shot rather than a video, you are asking the harder of the two questions, and the answer is more often "with the person who shot it" than "in a folder".
Why several locations is a retrieval problem more than a storage problem
Each place answers a different question. A drive answers what a file is called. An asset manager answers what somebody tagged it. A publishing platform answers what you shipped. None of them answers which video contains the shot of the factory floor, because nobody wrote that down at the time, and nobody was going to.
That gap is why the common version of this question is not really "where is the file". It is "how does anyone other than me get at this material", which is a different problem than storage and does not get solved by buying more of it. Teams often discover this after building the library: the files are in one place, and finding anything in them still means asking the one person who remembers.
What one home actually requires
A single location is maintenance, not a purchase. Someone has to upload, someone has to name things in a way the next person can guess, and someone has to own the rule when a project ends. The realistic target is not every asset in one place. It is one place that is reliably complete for the category of thing people keep asking for, usually finished brand videos and the interviews or event footage worth reusing. How that division of labour usually gets drawn is covered in the way teams split storage from retrieval.
Putting a copy somewhere indexed also changes the kind of question you can ask of it. Searching by description in Vivu returns time ranges from several different videos in one project, each with a line explaining the match, and some of those come back as near misses you keep or discard yourself rather than clean answers.
When the sprawl is fine
If the archive is a year old, if one or two people shot all of it, and if requests arrive a few times a quarter, spreading assets across a drive and a platform costs you very little. The tidy-up only pays for itself when people who were not involved in the original production start needing the material, or when the volume passes what one person can hold in their head. Before that point, a shared naming convention and a pinned document listing where each project lives does most of the work.
How to decide where to spend the effort
Listen to the shape of the requests you get. If people keep asking where a file is, the fix is one home and a naming rule, and it is cheap. If they keep asking which video has a particular moment in it, storage discipline will not help, because the information they want was never recorded anywhere. The first problem is solved by tidying. The second is solved by making the content itself searchable, or by accepting that the answer lives in someone's memory. Sorting your last twenty requests into those two piles tells you which one you actually have, and the hunt for one specific missing file is worked through in the single-file version of this search.
FAQ
Should finished videos and raw footage live in the same place?
Usually not, because they get used differently and have different lifespans. Finished videos are requested by people outside the production team and should sit wherever the widest group can reach them. Raw footage is bulky, only useful to whoever is editing, and needs a decision about how long you keep it. Splitting them is fine as long as one document records where each kind goes, and as long as somebody owns that document.
Who should own the brand video archive?
Someone whose job includes it, rather than whoever happens to be the most organised. In practice it is often a marketing operations or content role, because the person needs to care when a project ends without its files being filed. Ownership by a team rather than a named person is the arrangement that most reliably fails, since nothing gets refused when it arrives wrongly named.
Is it worth uploading footage from several years ago?
Only the part of it someone has asked for. Old footage usually loses value for a clear reason: the product changed, the office moved, the people in it left. A reasonable filter is whether you would be willing to publish a shot from it today. Anything that fails that test can stay wherever it is.
What if the only copy of an old video is with the agency that made it?
Start with the contract rather than the drive, since delivery and retention terms decide what you are entitled to ask for and how long they were obliged to keep it. Agencies often hold material longer than their obligation, so asking directly is worth doing even after several years. It is also the moment to set a delivery rule for future projects, because this question repeats otherwise.