If you do a little research you will find instances where the NSA or the CIA have pressured companies responsible for encryption technologies to include a backdoor for their use, or have simply (more recently discovered) bought said company to compromise said encryption. There are also encryption methods that are discovered to simply be weak, or have some flaw in their process that render the encryption useless. Sometimes lack of understanding by the user in how to properly implement the encryption, or protect their keys can be their undoing.
GPUs have greatly decreased the amount of time it takes to simply brute force encryption, so that something that might have taken days, weeks, months, or years, may only take hours especially if you take into account the use of clusters of GPUs. Government agencies have bought up playstations to be clustered for this very thing. Once while playing around with WPA cracking I noticed the difference between using my CPU vs using a CUDA enabled GPU was a time difference between a few days and 30 minutes using a dictionary attack.
Whether an encryption can be compromised depends on the technology available to the attacker, the amount of time they have available to perform the attack, and how important it is to them to break that encryption. There are some methods of encryption that haven’t yet been cracked as far as we know, but you must keep in mind that some times the encryption itself doesn’t have to be cracked, sometimes another flaw somewhere else in the system can be used to get access to whatever information has been encrypted.