I was surprised, but most gun crimes are committed by a small subset of people, and most multi-offend. I expected them to get their weapons from theft, but there's a surprisingly large (I think majority) which come from private party out of state sales -- at least in LA, a lot come from accomplices buying in AZ and then reselling. (for handguns, which are the vast majority -- there are some more exotic automatic/smg weapons brought from mexico, while US handguns go down to mexico)
There should be a way to catch the out of state buyers. A lot of the associates of criminals are themselves criminals, so NCIC on private party sales would actually push them toward more stolen weapons. If you could additionally increase the odds of catching these straw man purchasers and giving them felony convictions, you'd seriously hurt the supply. Combine that with safe storage and theft reporting, and you'd make a much bigger impact than the AWB, at a much lower cost to legitimate gun owners.
The question is whether pushing criminals toward theft would actually accomplish anything. If they still end up with largely the same number of guns, all you're doing is increasing the rate of theft, which could actually be more dangerous (especially where the criminals and the theft victims are both armed). But I guess we don't know unless we try.
I would also be interested to know whether the repeat offenders are largely gang members, because that probably changes the outcome quite a lot: A background check is a lot more likely to stop an individual felon than a gang with the manpower to knock off a gun store periodically or intimidate the owner into supplying a stream of firearms under false documents or even set up a machine shop and manufacture them internally.
The repeat offenders are mainly gang, mainly drug gang. Most gangs have some sort of hazy membership procedures compared to other organizations, so you've got a lot of affiliates. (the irony here is I know about US gangs primarily due to talking with FBI/police people who were working with the US military in Iraq/Afghanistan to teach them how to go after crime/terrorism/insurgency using police tactics...)
Robbing gun stores is actually a fairly unsuccessful crime. Outside MS-13, virtually none of the street gangs operating in the US has any degree of tactical proficiency; they can't go toe to toe with police patrol officers, let alone the kind of people who tend to own gun shops (in their own territory...)
Most of the robberies for guns seem to be burglary (and in places where people are morons and leave truck guns visible in their vehicle...), or might be faked somehow (like a way to transfer by calling it a "theft", but actually being compensated). They're not against gun stores usually.
In-house manufacture is definitely a thing (and ironically you can make a full-auto open-bolt weapon more easily than a normal gun), but not generally done as much in the US due to easier availability of other guns, and brand/hollywood preference for specific types of pretty guns.
Restricting the sale of things doesn't do anything about their installed base. Gun owners already own guns. They can currently do or obtain the thing you want to ban -- and every time you even mention banning something, the sales go through the roof. So, as a gun control advocate, you might want to stop doing that unless you're sufficiently positive both that you'll succeed and that your success will outweigh the instantaneous increase it causes in the installed base of the bad thing. Which is a hard calculus to accept, because of the installed base, which makes any dividends that come from a ban only accrue very slowly over a long period of time. You aren't causing certain classes of firearms to not exist anymore, or taking them away from the classes of people who already have them, you're only causing there not to be a single digit percentage more of them next year. Which is pretty weak consolation if the discussion of the ban caused there to be a large multiple of that increase this year, and most of the money that would have gone to the increase in future years just goes to the nearest analog or a purchase by proxy in the future years.
Moreover, you can't identify any subset of guns that can be used to kill people and ban only those. You can kill someone just as dead just as fast with a revolver as with an AR15, and the differences between an AR15 and a hunting rifle are entirely cosmetic.
And prohibiting high capacity magazines is only even theoretically productive if your sole concern is mass shootings, which are such a small percentage of gun violence that it's very hard to take seriously as a meaningful reform.
The problem is that real gun control, i.e. going into peoples homes and taking away all their existing firearms, is comprehensively politically infeasible even regardless of the constitutionality.
So we get half measures that are at best ineffective. Bans on guns that look scary. "Background checks" even though first time criminals won't be flagged and hardened criminals are more than willing to steal or pass a false ID or use an accomplice with a clean record. They're all just security theater to satisfy the need for something to be done.
And though I am not a fan of guns, I find myself quite sympathetic to the argument that the proper solution is to go to the root cause and do what is necessary to reduce the number of people who come to find murder as an acceptable solution to their problems, rather than going out and trying to prohibit every dangerous thing an already-malicious actor could use to harm another living soul.