Despite all this, nobody seems to claim that cryptography is 'futile'. Sometimes mere infeasibility is enough.
Modern DRM is typically in the class where you can either spend at most a few days with a debugger and a code editor, or buy a $20 knockoff device to break it.
These are not really comparable.
Claims like these are often misleading as they assume that i) malicious parties' capabilities, ii) what is and isn't possible, and iii) implementation correctness, are all static known factors. Cryptography is more akin to putting a lock on a fence - it's going to keep most people out or perhaps deter prying eyes, but it's all probabilistic in the end and depends on who's doing the prying.
It is not a question of theory, but practice, DRM gets systematically broken all the time because of this.
With DRM it's usually the other way around. The attacker always has the encrypted data, the key and decryption routine. Otherwise playback wouldn't work. The only defenses are obfuscation and anti-debugging, which are basically security by obscurity. As soon as someone understands your scheme it's broken for good and you have to come up with something new.
Just like with cryptography you're trying to make it as hard as possible for attackers, but with DRM it will always be around the same magnitude of difficulty as for yourself.
Crypto OTOH can be completely impractical to break. Plenty of codes around that would take even the NSA millions of years to crack, baring new mathematical or computing (e.g. quantum) breakthroughs.