It is very common to claim that this or that armed service needs to buy x or y simply because the specialised industrial capacities would otherwise break away, with a terrible loss of industrial competence. This is often an argument behind portions of military budgets that remind rather of industry subsidies than actual military procurement, such as the third buy of a ridiculously overpriced EGV (Einsatzgruppenversorger, fleet supply ship) for the German navy.
Several facts irritate me in this regard:
Many suppliers may have some competence, but often times their corporate culture is so rotten and wasteful that creative destruction might be worth it. Shipyards and generally arms industries with a reputation for producing overpriced products are prime suspects.
Nobody can prove that civilian industries are unable to convert to military products in a matter of a few years. Actually, the World War experiences suggest the opposite.
Maybe it takes technical expertise at the procurement agency to retain the necessary competence, not in the industry? Civilian industries are used to produce according to their customer's specifications. The ability to specify what you want is probably the key, not keeping an industry used to your specifications.
Historical examples. The French aviation industry got many small orders throughout the 20's and 30's from the Armée de l'air, but it was worse than the German one by about 1938 (there had been almost no German aviation industry during the 20's!).
The French were late to convert from old airframe technologies to all-metal airframes, did not innovate much and their production capacity was rated as manual and unsuitable for quantity production by foreign visitors.
Lobbyism. There's a systematic bias in favour of the status quo industries not only because of widespread intellectual laziness and risk aversion, but also because said existing industries can spend on lobbyism. Not-yet-existing suppliers cannot do the same.
Large, modern procurement projects last for so long and are so expensive that it's difficult to imagine that civilian companies which survived the fierce competition in civilian markets and met the requirements of civilian customers with projects lengths like one to six years could perform worse.
The only good explanation for such a scenario seems to be the A400M pattern; a civilian industry might misunderstand a military development & procurement contract for a subsidy.
I wished that this "we need to retain the industrial competence and capacity" argument would be taken with a grain of salt, not as gospel. It really doesn't deserve to be treated as "black box" argument that requires no further inquiry. It's likely more often than not a very, very poor argument.
edit 2010-08-20: Changed 1936 to 1938.