Stakeholder demands (not accepting a bland and functional station) are the main cost driver, although architects often design blind to the budget.
This is a popular view but it's an oversimplification of what actually happens in most projects.
Most projects start out with a set of aims, and a target budget. Quite often these two things are not compatible; that is the desired aims can't realistically be achieved within the desired budget. If that's obvious to the architect at the outset, then of course they ought to say to the client. The choice then is either to scale down the brief or accept that the budget needs to be raised. Clients commonly don't want to do either, and ask for design work to proceed based on the initial brief anyway, hoping that savings can somehow be found later. It's rarely in an architect's interests to do this, because it just means there's likely to be a lot of abortive design work, for stuff that later needs to be omitted for cost reasons.
Sometimes, it's just not really possible for an architect to judge costs at the outset of a project. There are all sorts of reasons for this. Maybe the brief itself simply isn't well defined. Often you have to go some way into the design process to work out what the brief actually needs to be. Maybe you are operating during a period where material or labour costs are changing an unpredictable. That's been the case for the past few years. Maybe the high-cost elements of the build aren't really related to architectural decisions as such and are more to do with engineering requirements, some of which might not be known until the site has been investigated more fully. There seems to have been some element of this at Euston.
The idea that architects (or other consultants such as civil or structural engineers) have walked away laughing from the Euston scheme with these millions of pounds of fees is a bit silly. Those fees are payment for work that has been done. Probably hundreds of people will have invested many hours of their lives working up designs in good faith, only to find that decisions made by the "client", in this case ultimately the DfT, make all that work redundant. It's pretty common in the building industry to be working on designs that never get built, because projects get changed or aborted for all sorts of reasons. No-one really wants to be doing abortive work. Most designers want to do their job well and want to see a project come into reality and be successful. The image of some kind of "gravy train" where all these people sit around doing unrealistic designs while money is thrown at them is kind of insulting.
Reading the NAO report, there doesn't seem to be any indication that the problems have originated in the design team drawing up clearly unrealistic or gold-plated schemes. The issues seem mainly to do with the various changes of mind at a much higher level.
Also, I don't have any detailed knowledge of the various designs that have been proposed or abandoned, but my guess is that cosmetic architectural stuff is not going to form a large proportion of the cost of this scheme. My guess is that the costliness is mostly going to be related to the massive engineering complexities of that site. Not just the basic structural issues of building around existing structures but the sequencing and logistics of the building process too.