Sonic.net has a different cost model for transit
The transit cost model for small regional ISPs is very different than large national networks like TWC and Comcast. It's the basic build vs buy scenario that comes about with scale.
Sonic.net pays carriers for transit so that they will deliver bits to them in California. No matter where the traffic originates, carriers must carry that traffic out to California to hand it off to Sonic.net's network.
Comcast and TWC have national footprints, so it makes economic sense to build out their own national backbone. This also enables them to negotiate better transit pricing by taking on traffic in regions close to the source. If a TWC customer in California starts pulling video from a server in Atlanta on nLayer's network, nLayer can dump that traffic directly to TWC in Atlanta and TWC handles transporting that traffic back to their customer in California.
Here's where it breaks down: say Netflix is paying Level(3) for delivering content to TWC today. If TWC dedicates connections for OpenConnect, they will substantially reduce the traffic that transverses Level(3) to reach them. This reduction in traffic will lead to a decrease in revenue for Level(3) from Netflix, which will in turn cause an increase in rates for what Level(3) will charge TWC for bandwidth. (less of a volume discount, and have to make up for some of the lost Netflix revenue) TWC has to tie up network ports that are only used for a single Internet service (Netflix), and they will incur higher last-mile costs once they get this established because the bitrates offered via OpenConnect are higher than the standard service.
Most likely this will result in increased costs for a large national ISP, and those costs would be passed down to all ISP customers regardless if they are Netflix customers or not. Netflix, however, sees a huge financial benefit from the reduced transit costs. When you're dealing with $7.99/mo accounts, I'm not sure you can see enough of a cost drop to offset how much broadband rates would go up.
While your partially right, the truth is that the bandwidth used in all of those peering agreements will drop considerably, netflix accounts for a huge amount of traffic, especially at peak times.
even if level 3 raises rates, the isp is going to be requesting a lot less traffic, overall costs shouldn't increase for the ISP, and likely will decrease, though not as much as they will for Netflix.
The real loser in this scenario is level 3, I doubt they can seriously raise prices to the point it will make up for the lost data transfer netflix and the isp are using.