I am reading Elements of Information Theory and Theorem 7.13.1(Source-channel coding theorem) basically says that we cannot use a channel if the entropy of our source is higher than the channel capacity. To me this does not make sense intuitively.
For example let's say we have a source with entropy $H(X) = 2$ bits and a channel with capacity $1$ bit. This means we can send $1$ bit per use of the channel. Why does this have anything to do with the entropy of our source? My understanding is that we can just use the channel two times to send the message. This will obviously not increase the capacity(Lemma 7.9.2) but it will still allow us to send the source over the channel.
What am I missing?
The key here is to understand that the (fundamental) information theoretic quantities are rates, whose values correspond to a ratio of number of "information" units (bits) to the unit of "time". For example, $H(X)=2$ does not mean that the source $X$ generates $2$ bits, it means that it generates 2 bits per unit of time and does so for an arbitrarily long time interval.
Now, the goal in the source-channel coding theorem is for the receiver to be able to (reliably) reproduce the source (located at the transmitter side) at the same rate as the information is generated. A delay in reproducing the source is allowed, however, once the decoder starts producing output, the output must be at the same rate as the source. In order to do so, under a discrete memoryless channel transmission, it is necessary to hold $C>H(X)$. And this is intuitive: As the source generates $H(X)$ bits per unit of time, the channel should be able to support the transmission of at least $H(X)$ bits per unit of time (or, as commonly stated, "channel use").
In a "real-world" setting, the theorem states that you could not stream an HD-quality movie using the internet infrastructure of 30 years ago, simply because the link capacity could not support this high rate of information. Of course, the approach you suggest would eventually be able to transmit the whole movie, however, it would probably take a long time to do so!