No, it's not exact to say hyper(hypo)methylation causes cancer.
As you may know the current hypothesis of cancer progression assume it requires several (5-6) hits that have to accumulate to make a malignant metastazing tumor (this one is pretty good review Hallmarks of cancer).
Each of the hit is though to be an acquired (doesn't to apply to hereditary cancer syndromes) mutation that enables/disables some genes important for the cancer features.
But epigenetic regulation is also a way how to enable/disable gene. For instance a specific methylation pattern is characteristic for stem cells. Cancer cells have many properties of a stem cell, like selfrenewal and unlimited replication capacity.
Epigenetic changes help cancer cells to acquire these features more easily.
The epigenetic pattern in cancer cell is different, tumor suppressors are hypermethylated to be disabled, but some other genes may be beneficial in hypomethylated status (like those creating a pseudo-stem cell signature).
It's tru that demethylating agents have been used/tried in cancer treatment, because they expect the tumor suppresors to be activated which would be enough for the cell to die, but it can't be said, that hypermethylation is the problem. And, those demethylating agents have a questionable efficiency.
So, methylation is a way how to swich genes quickly on and off and in targeted fashion (much qicker than to wait for a random mutation and then select that clone), and cancer cells have altered methylation. Because it is a tool to be used in many ways and cancer cells use them to promote their survival and replication.
The more flexible methylation control (instead of the classic one that governs for example commitment to a certain differentiated lineage) is proposedly the reason (or maybe one of reasons) of the observed cancer cells plasticity and genomic instability. It's quicker and switchable, unlike DNA mutation.
Here is another picture from a very nice review.