Claim Proprietary Wolfram Language is 5-10x faster at coding for compute intensive jobs

There are some fascinating claims here. They point out there is general search (Google) compute search (Wolfram) and machine learning search (IBM Watson.) They claim they beat IBM Watson a lot of the time because its hard to apply for companies application sets. They claim Google’s approach is 70-80% tapped out. They claim their compute approach by contrast is only 10% tapped out on its potential.

I want to know what people would think of the fusion of the approaches. Seems inevitable as part of the AI picture.