[robmccoll]

For most developers, it doesn't. They're both making x86_64 chips with coherent caches of similar size, similar core and thread counts, and a lot of overlap in instruction set extensions. For people doing very high performance work (heavy data processing, simulations, etc.) or performance critical work (think hand coding optimized crypto library routines in assembly), the subtle differences in those categories might affect how they lay out data and access it, how many threads they use and how they use them, and preferences for the availability of instruction set extensions that are specific to the kind of workloads they have (vector processing extensions, native crypto operations, unusual bit twiddling patterns).

[sanxiyn]

Another difference is that Intel and AMD virtualization (vmx and svm) are incompatible with each other.

[hitpointdrew]

>For most developers, it doesn't. They're both making x86_64 chips

This is true for the vast majority, but there are niche cases were there are differences even though both are both x86_64(like Intel's FlexMigration and AMD-V Extended Migration).