By any means, I'm not saying the PS4 doesn't have more of a GPU, because it does. The thing is though, it needs that GPU when you've got a CPU crippled by GDDR latency. Audio processing (not be confused by the audio encoder in the PS4) will have to be off-loaded to the GPU, a lot of the physics will be handled by the GPU. Those extra CU's start decreasing and decreasing and when you've got a CPU which you have to think a lot about because they've put GDDR in there, then you're starting to see what Albert Penello is saying.
-CPU isn't crippled at all by GDDR5 latency. Their hUMA architeture and bussing should mitigate any large concerns about that.
-Microsoft's specialized audio hardware is also for encoding, the additional hardware is for Kinect's special purpose processing. Xbox One still needs to offload audio tasks to either CPU or GPU, so this isn't some advantage you're making it out to be.
-Physics should be handled by the GPU thanks to GPGPU in both Xbox One and PS4. Again, this isn't an advantage for Microsoft, in fact the contrary; PS4 has far more hardware dedicated to GPGPU so expect to see larger differences as developers come to grips with asynchronous compute.
-Those extra CU's are going to get plenty of work. Asynchronous compute allows developers to get extra performance out of idle GPU performance.
-Albert Penello is relaying information (or misinformation) about their own hardware, and isn't saying anything about the PS4's hardware choices. I'd love to hear your reasoning about why his word is better than Mark Cerny's.