Since high quality, near-lossless 48kHz mono audio can be encoded at about 96kbps, which is much lower than most users' upload/download bandwidth, why do provid

Since high quality, near-lossless 48kHz mono audio can be encoded at about 96kbps, which is much lower than most users' upload/download bandwidth, why do providers even think about going under this bitrate? For example, Skype is ~30kbps and frequently sounds overly-bandlimited and garbled, so why doesn't their algorithm just use only three times this amount? Is it an issue with latency, or just keeping down their server costs? I legitimately want to know why, I'm not ranting.