The optic nerve is a bundle of nerves like several other nerve bundles in the body. I suspect that it has more nerves than any other simply because it must convey a few million pixels. I have heard of no suggestion that a single axon in the optic carries information for more than one pixel. For most and perhaps all purposes of the optic nerve adjacency of pixels is necessary to achieve from one end to the other. Nerves that connect various visual processing areas in the brain have even more complex integrity requirements, but perhaps less stringent. The latter are shorter too which helps.

I know two techniques to wire up the optic nerve and neither seems highly plausible:

This suggestion posits that the proto-retina and V1 within the cortex develop together into a 2D space and then separate but always keep a bundle of nerves between them.
About 30 years ago someone proposed an image encrypting technique that would start with a glass fiber bundle with enough fibers to transmit the image over a few feet. Then the middle of the bundle would be processed to disassociate the fibers from each other so that they could be scrambled. The two ends of the bundle would be held firmly in this process so that the scramble bundle could still transmit an image. Next the scrambled middle of the bundle would re-congealed and cut. The two halves of the bundle would serve to encrypt and decrypt an image. In short the pixel locations are scrambled.

The cryptanalysis of this system assumes many encrypted images and looks for pairs of pixel locations that agree more often than most pairs. This suggests that such pairs occurred close to each in the plain image. Enough such discoveries will piece together the whole picture. The same algorithm might conceivably work for the optic nerve.

This speaks of the morphogenesis of the eye, but does not speak of the optic nerve. It covers some if the difficult problems with a new eye and the history of attempts to solve those problems.