This is about blob storage not 'AWS' and 'Azure' or 'HP'
IMHO the basis of the study is invalid. But before writing any more, this comment from the report seems relevant:
"In addition, this distribution of file sizes closely matches a well-documented breakdown from a study conducted by University of Wisconsin and Microsoft Research2."
Scope for bias?
This report is not about 'AWS' or 'Azure' (no VM performance, no database performance comparison, no real application stuff). This report is about comparing the blob storages. In the case of AWS, S3. S3 is great but it's primarily static file storage. Plus as a front end it evaluates access policies. If I want performance to store blobs, say as part of a transaction based application, S3 is not my first option and, though maybe I'm wrong, I don't think it wouldn't be the AWS recommendation either, especially for small files.
Oh, and the OP didn't mention that when considering files > 1MB AWS is fastest. 70% of the files in the study were 100K or less. 9.7% were 1M and 22.2% were 10MB. Analysis of files we store in S3 shows that over 90% are more than 1MB. The study referred to looks at files on the PCs of Microsoft employees. Are the file sizes on local PCs representative of the files that will be stored in a blob storage? Not in our experience.
Before choosing AWS we looked at both services and found Azure to be more expensive on a like-for-like basis (insofar as there can be like for like). I admit that read/write performance to S3 was not on the list of things to consider. We use S3 to store and deliver videos and similar. The variability of user experience accessing these file is more likely due the vagaries of the internet than any characteristic of the storage infrastructure. Which, by the way, AWS provides its customers the change to ameliorate by using Cloudfront so the content is delivered from (relatively) local edge points. There's no mention of the impact of edge location services in the report.