If that argument holds, it effectively outlines the product spec for open-source distributed compute, where many groups, individuals, and organizations pool compute for training large-parameter open models