On starting things before there is a business case
There is a respectable position that says nothing should be started without a business case, and there is a less respectable position that says business cases are a failure of nerve. Both are wrong in the same way: they treat "is there a case" as one question when it is two.
Two kinds of unknown
Some unknowns are answerable in advance and cheap to answer. What does the equipment cost, how long is the permitting queue, what is the addressable volume, who else is doing it. Refusing to answer those before starting is not courage, it is negligence, and it is expensive in a completely predictable way.
Other unknowns are only answerable by doing. Whether anyone will turn up. Whether the two organisations can actually work together once the memorandum stops being the main artefact. Whether the thing is any good. No amount of modelling produces those answers; it produces confident-looking ranges that turn out to have been arguments about assumptions.
If the number you need can only be discovered, modelling it is a way of deferring the discovery at full cost.
The actual test
So the useful question before starting is not "is there a business case". It is: which of the unknowns here are of the second kind, and is being wrong about them survivable?
If the discoverable unknowns are the important ones, and being wrong costs an evening or a quarter rather than a decade and a reputation, then starting is the cheapest available research method and the business case is something you write in six months with real numbers in it.
If being wrong is not survivable — if it takes a factory, twelve years and other people's pensions — then the modelling is not cowardice. It is the job.
What this is not
This is not permission to skip the work. Almost everything that gets described as "just starting" is a refusal to answer questions that were entirely answerable, by people who found the answering boring.
The distinction is narrow and it is the whole of the method: answer everything that can be answered, then start on the rest rather than pretending it can be answered too.