Given the constraints in your question, yes, it can be predicted.
For symmetric-key encryption algorithms, the answer is "never".
Assuming Moore's Law is valid for the next century (an extremely optimistic assumption -- such a computer would be drawing much of the energy output of the Sun to power itself) and that current computers can test a million keys a second (a reasonable assumption), a computer in 2114 will be able to test 7.3*10^25 keys/sec. At that rate, it will take approximately 146,000 years to brute-force a 128-bit symmetric cypher (the minimum considered "secure" right now).
Asymmetric cyphers are a different matter. Since they depend on certain mathematical operations (eg. computing discrete logs or factoring numbers into large primes) being difficult, they're vulnerable to improvements in the algorithms for performing those operations. Improvements in computing power are relatively easy to predict; mathematical breakthroughs are rather less so.