100 mg Silver (1 liter 100 ppm; 250 ml 400 ppm, etc) / 107.87 x 20 000 (or 100 000) = 18540 mg to 92704 mg, or 18 to 93 gram. And if we assume that only 10% of the silver atoms are at the surface of each nano particle in average, then 1 / 10 of that, or 1.8 to 9.3 gram. I still believe that this is way too much of gelatin.

In this paper, their best results was a weight to weight ratio of gelatin to gold of 31. At that ratio for gold, the equivalent ratio for silver would be close to 60. So for 100 mg of silver you would need 6 grams of gelatin, which is within the range you just calculated. So your estimate is not unreasonable, and the manufacturer does not disclose the molecular weight of the gelatin on their website.

To prevent 'jello', this would put a fairly low limit on the ppm which could be achieved. I use 1 gram of Knox gelatin for 80 mg of silver in 250ml of water which is a lower ratio than above. Much more gelatin than that and it starts to set up and become quite 'gloppy'.

Note that the physical size of a gold atom is only 5% bigger in diameter than a silver atom even though the gold weighs almost twice as much.

You can estimate the number of atoms in a nanoparticle and the number of surface atoms by knowing the diameter of a silver atom is 330 picometers and the packing ratio for spheres is 75%, and assuming the atoms are touching each other. This is where I got the 10% number.