Alright, perhaps I shouldn't have quoted that, but let's continue with the 2 edits of my post where I found real figures on revenue and costs. Video revenue is (in absolute terms) higher than internet revenue. Video revenue growth is slower than internet revenue growth. Video customer growth is actually shrinkage, they're losing customers (hundreds of thousands each year). Internet customer growth is actually growth (over one million each year).

The primary video cost (programming) is approaching 50% of video revenue, and it's growing at a pace that has literally equaled or exceeded video revenue growth for each annual period in the linked document. So, making an assumption here that's likely untrue, if we consider that the other costs are distributed equally between the two (marketing, tech support, physical support, etc) then the internet service is approaching parity with the video service, and will exceed it in several more years as the higher profit business component.

I suppose if someone else has more time to dig through this document or other insight from being in the industry they could chime in with which has the greater costs once programming is removed from the equation.

The problem is that you can't model the ISP portion as something with fixed costs to build and then small marginal costs of moving bits around. Building it out is the primary cost to the ISP portion and we, the customers, rightly demand they keep on doing it. They have to always keep on building more capacity. You can't say "okay, now they have stopped building capacity, it's all rent-seeking from here."

The actual electrical cost to move the bits back and forth is nil. It's making sure that half the people in the neighborhood can use Netflix at 7pm two years from tomorrow that requires them to spend money today.