I think it would be very, very hard. You would need to provide something that existing for-profits don't do (open access), while also competing with non-profit

I think it would be very, very hard. You would need to provide something that existing for-profits don't do (open access), while also competing with non-profits that have the same goal. The problem is that scientists who care about open access are not likely to move their work to another for-profit enterprise. There are plenty of non-profit entities, from universities to professional societies to PLoS to the government, who have an interest in making research open, and don't need a profit (though they do need operating revenue). It's a tough sell to make your for-profit business the one that a scientist would submit papers to, or review papers for, when they could choose either an established for-profit with higher impact, or a non-profit.