Having performance issues with just 400k triangles in mobile signals something is not correct, mobile can do a few million vertices smoothly if done right.
The biggest issue on a mobile's GPU is memory bandwidth, which can explain your issue depending on how you are rendering those 400k triangles.
Batching (combining those 100 objects into one) won't solve the problem here, as GPU still has to go through all those vertices because it doesn't know they are just a copy of the 1 of 100.
What you want is to tell the GPU that you have 100 copies of the same vertex buffer, that's called Geometry Instancing (or just instancing), and it's largely supported by mobile devices nowadays. Keep in mind that for this you will need its own shader, I recommend also you calculate your object position on the fly, as sending 100 matrices to the shader in mobile may not be as fast as simply procedurally positioning them yourself.