I respectfully disagree with that author. In fact, our release directly falsifies this claim:
Think people, if it was possible to come up with a next generation SHA256 or Cryptonight or whatever miner that could possibly be profitable to warrant a redesign of what is already out there
then Bitmain would be shipping the S10 today (which does not exist)."
The assumption that BitMain is the best at everything is simply false. Currently, WhatsMiner is significantly better, for example.
Also, consider the community-owned / blockchain-subsidized model. If, for example, 20% of the mining reward goes to subsidize a community-owned manufacturer that sells units at-cost, then any for-profit company couldn't really compete. You can see this with shipping companies in the US. UPS and FedEx have to offer special services at large price increases compared to the subsidized USPS. There is no way UPS or FedEx could ever deliver a letter for 50 cents. It could be the same with a mining company.
In the case of CryptoNight specifically, we have already open-sourced a design which knocks the socks off the leading competition. The best BitMain could do is copy our design.