Let's say you have two containers of fluid (in picrel I draw them as open, but say they're sealed except for a "door" or "aperture") billions of light-years apart, such that the expansion of space between them is close to the speed of light. Set them on a constant velocity that exactly counteracts this expansion such that they remain fixed relative to each other. Now, perform the following process to create energy:

On one end, set a projectile moving toward the other fluid container at some arbitrarily small but non-zero speed. As it moves toward the other fluid container, it picks up speed due to the expansion of space, reaching a top speed of the expansion rate of space between the containers, i.e. arbitrarily close to lightspeed. This means that when it arrives, it has an arbitrarily large kinetic energy, based on how far apart the fluid containers are. Briefly open the small opening in the fluid container mentioned earlier to allow the projectile to enter. As the projectile moves through the fluid it heats it up. Gather up this heat as usable energy. Then, to reset the process, use some energy to compensate for net movement of the fluid container and return to its original location relative to the other. Then with the leftover energy, use energy to matter conversion (if the projectile speed exceeds about 0.9c, the kinetic energy exceeds the rest mass-energy of the projectile) to replace whatever fluid leaked out when the opening was opened, and to create another projectile. Send that back to the other fluid container and repeat the process.

I'm posting this under the assumption that there's a reason it won't work, but that I don't know what it is. I feel like there might not be enough energy to reset the velocity imparted on the container by the projectile, but I'm not sure if that's the case, or if there might be another reason it fails.