https://en.wikipedia.org/wiki/Osmotic_power
The power density even at 40% efficiency is pretty high (300Wh/m3) so major rivers (Mississippi 20,000m3/sec or 72Mm3/hr) would generate 20GW fully exploited.
The problem has always been the growth of biofilms on the surfaces of the membrane, which break them down and/or lower their efficiency.
Also, as always, curious how having these membranes in the water would affect the local ecosystem.
Besides, it's not just biofilms that are the problem. With a river you're going to have huge amounts of silt and very fine clay clogging up your membrane. The maintenance costs for an installation like this would be overwhelming.
[1] https://www.eia.gov/realtime_grid/#/summary/demand?end=20160...
And the output is just mild salinity water, so you could probably extract form 1/2 the flow without causing that much environmental harm as long as you pipe the output to an area with similar salinity.
Still a pretty cool thing if they can just install these in a river mouth and get free electricity out of 'em.
[1] And an external heat source such as a coal fire.