> Also I’d bet money ARM has multiple in-house processors for testing/R&D
When I worked at ARM, we just used FPGAs to emulate the chips and waited for the customers to give us a board with the actual chips. Maybe another division did create their own hardware, but everything we worked with for testing in ours was from customers.
> there is a big difference between the generic/basic patents we’ve seen trolls use over and over again (stuff like “an input device causes and output device to do something”)
If it is a generic/basic patent, why should it make a difference whether the enforcer is a practising entity (PE) or not? Unlike a non-practising entity (NPE), a PE would be less inclined to licence it because they want to keep competitors out. ARM is big enough that they can instigate litigation on their own against infringers. Smaller inventors have no chance of doing that, and patent trolls are often their only venue. Many times, the problem is with bogus patents that should never have been granted because of prior art or obviousness (supported by the authorities failing to perform due diligence before granting patents). And the other is the expense of defending against a valid patent that does not apply to the defendant's product. Patent trolls' portfolios are often full of such patents, but this is not limited to NPEs.
> big companies will just straight up copy or buy up any competition
An example that comes to my mind is Nuance, a practising entity that engaged in frivolous patent infringement lawsuits against a new competitor Vlingo. Nuance did turn their patents into products (Dragon NaturallySpeaking), so they are not considered a patent troll under the usual definition. Yet, their actions against Vlingo were the same as that of a troll, and even though Vlingo won, they were weakened enough to be bought out by Nuance.