Purpose: To display an image on my screen given an image description that was obtained via HTTP from a public, potentially eavesdropped, Internet connection.
Problems: Security - every browser has had exploits with information disclosure or, worse, some even have arbitrary code execution flaws.
Solution: Convert the already overly-complicated array of several protocols and image formats that do nothing more than describe an image, which you already render in the worst possible way in non-compliance of all published standards, into a format suitable for GPU processing via hardware accelerated interfaces involving yet-more protocols and conversions, reliant on driver compatibility and hardware being available, and on only a single operating system that has those interfaces (renowned as the most attacked system in the world, usually via browser weaknesses), in order to render the 1Mb page that took several seconds to decrypt/authenticate/respond/download onto the screen a teensy fraction of a second faster, consuming more resources, generating more heat and using up more of your computer's bus bandwidth than strictly necessary.
Well done, MS. You've got that one aced. Can't imagine why no-one else wants to follow your lead especially when, on the whole, their browsers already match or beat this new one on almost every count (security, stability, speed, standardisation).
Now if you'd said that the browser loaded quicker, I'd actually be more interested (something tells me the startup times might be ENORMOUS even if the render times are technically quicker compared to its predecessor). Or worked on XP (acceleration or not). Or had had no security problems in several years of beta. Then I'd be impressed.
Didn't MS spend years trying to prove that IE *was* an integrated part of Windows, only to have it proven that it wasn't, only to be told that they had to separate it, to now making it as vastly Windows-only as they could manage to the point that it doesn't even work on some versions of Windows any more?