I think the point that ESR is trying to make is that it doesn't matter if some company steals open source software and uses it to create a proprietary product. ESR believes that the advantages that FOSS have over proprietary software are so great that companies that engage in that kind of activity still won't be able to compete against FOSS solutions.
I'm not sure I'm ready to agree with ESR, but he's probably spent a lot more time thinking about it than I have.
Microsoft hates the GPL because it prevents them from using their favorite tactic, Embrace and Extend, to try to crush FOSS. What MS doesn't realize is that it's too late for even those kinds of moves. As ESR points out, the standards bodies have caught on to the kinds of things that MS does and are not relying more on reference implementations of standards which will block MS from taking over critical protocols.
More and more businesses and governments are realizing that FOSS just makes sense. It's not the GPL which is selling them, it's the total cost of ownership numbers. I think many of them also value the ability to customize the software as needed. Imagine what it is like if you are a small country that uses a language not supported by MS or other proprietary vendors. Your best chance for having office software that you can use in your native language is FOSS.
The bottom line, if I understand ESR correctly, is that the lowered process friction of FOSS means that the open source community will always be able to out innovate any proprietary company through sheer force of numbers. While this idea works well for popular FOSS projects like Open Office, Linux, and Apache, I'm not so sure it's true for every FOSS project. I think I would agree that the GPL might not be required on those projects, but it might still be important for smaller projects that don't generate so much interest that their developer base dwarfs that of any proprietary company.