Many technical reasons - I actually had the misfortune of a job which required use of Azure for 6 months.
The disk performance is awful, the instance spin up time is terrible, the management tools are either non-existant or significantly behind where other cloud providers are - a common response from priority support is “click here in the management portal”. It also has the legacy of being a half-assed IAAS bolted onto a half-assed PAAS, with things like “Cloud Services” which don’t fit any form of sane mental model, and very poor virtual network support both in terms of feature set and management tools (there is no concurrency control on the management API for virtual networks, for example).
There is some good - their object store isn’t bad and is quite cheap and I’d be tempted to look into websites to host our commercial debian repositories since S3 doesn’t support basic authentication though I dislike running on IIS immensely and it still supports insecure ciphers.
So in summary, if we ever do an IAAS and I’m involved in configuring or maintaining it, it will be in AWS (incidentally, all our build system, Windows development boxes etc are there too).