Great idea: hurry up and send me your money so that I can develop a new language and license it for any “commercial” (according to Will, no doubt) use. It will produce “better benchmarks,” yes. And, tell us all about it, Will, what about the hundreds of thousands of dollars a year in royalties that you dream such a contraption, funded by others, would produce for those of you who “own” it? Your apparent motives in this are not well hid.
Unfortunately, language compilers are notoriously bad revenue producers. There are simply too many of them out there already which produce excellent runtime results (when you actually need such results), and which are not only royalty-free but free. (Like gcc, for instance, or perl.)
Runtime speed is easily dealt with: you throw Moore’s Law at it. Even if the machinery cost five-figures in the currency unit of your choice, that pales in comparison with the cost of rewriting it ... or of trusting your soul to a brand-new language system.