No sequencing machine reads a whole chromosome from end to end in one pass. Instead it reads many short fragments, each only a few hundred letters long, taken from different places along the DNA. The output of a sequencing run is therefore a large pile of short reads, not one tidy sequence.
To recover the original sequence, the reads are assembled. The key observation is that the reads overlap: because the fragments were taken from random positions, the end of one read usually matches the beginning of another. Assembly finds those overlaps and chains the reads together, like reconstructing a sentence from many overlapping slips of paper that each contain only a few words. Where reads disagree, the assembler has to decide which letter is correct, and regions that no read covers remain gaps.
The result is a consensus sequence: one continuous string of letters representing the DNA that was sequenced. This assembled sequence, not the raw pile of reads, is what gets stored and shared as a genome.