Chemists and biologists can build medium-sized molecules and structures in a variety of shapes and materials.
It's hard to build large (greater than micron scale) molecular structures, especially large strong structures.
There are ways of fastening together molecules more strongly than DNA binding, but there's no obvious chemical way of controlling what fastens to what. Dissolved in solution, anything could bond to anything.
If medium-sized molecules were held in constrained position and orientation relative to each other, so that only desired bond sites could get close enough, then bonds could be formed only where desired.
A few years ago, Drexler (IIRC) described two ways of fastening molecules together. One is a protein structure called a zinc finger, wherein several amino acids (cysteine and/or histidine) bind to a single zinc ion. Another is a pair of paired carbon atoms, which rearrange bonds when hit by light to form a link across the pair. There are many other ways of fastening molecules, as well.
So, if DNA-tagged molecular shapes (whether made of DNA with Rothemund staples, or Schafmeister polymers, or whatever) were allowed to self-assemble to a DNA framework, and then zinc or light (or whatever reagent) were added, then the shapes could bond quite strongly into a single large strong precise molecule. The molecule could be highly crosslinked, and thus stronger and stiffer than most protein, and certainly stronger and stiffer than just plain DNA.
How big could the molecule be? Well, let's not forget IBM's recent announcement of plans to template entire computer chips with surface-attached DNA. That implies that the molecule - a single, engineered molecule - could be centimeter-scale!