Suppose that "cold fusion" is possible.
Handwave: look at, for example, Seth Lloyd's talk on quantum mechanics in living material, especially how the chloroplasts route energy to the active clorophyl with far better efficiency than would be normal, but the complex implements a quantum search protocol or something like that.
Look at catalasis in general. Lots of that in videos from SLAC (fuel cell and battery research) including an interesting talk on the role of metal atoms in biological molecules such as chlorophyl and hemoglobin.
Look at how smell uses electric bond resonance frequency.
Now...
Cold fusion would require a proton to be placed in a specific energy level in a specific quantum state. Photosynthesis generates protons that move around, and the antennas use careful quantum states.
So, such a molecular nanomachine could be built to do cold fusion in an organelle or small bacteria.
Problem is, the energy released probably destroys the system and everything in a small radius. The released energy is too high to be carried by phonons or bond vibrations, and the emitted neutron isn't easy to catch and turn the kenetic energy into power. So the energy release simply heats water or other working fluid (maybe liquid metal)
The bacteria or equivalent organells in a eucaryote cell could produce one-time-use fusion event units, and these are harvested in a concentrate. To use effectively, you isolate one in a droplet of water and set it off, producing superheated steam. The loose neutron decays after 11 minutes or so and deposits more energy in the water. (A volume of water is used to safely stop the neutron)
Early crude use would get more than one unit in a diluted drop thus wasting them. Higher technology would make use of each one indivually.
Maybe you need to use heavy water to build the nanounit. But the usage can breed more, recovering some percentage. The whole living form could locate and harvest the duterion from its minor contribution in plain water. A molecule containing duterium in place of hydrogen can be smelled (at least by fruit flys bread for the purpose) so they can be sorted out.