From a practical POV having all the training data, training infrastructure, and training know-how wouldn't help you either unless you could afford to spend the millions of dollars (hundreds of millions for a SOTA model) in compute to train it each time they released a new training set, in which case you're only talking about the big commercial companies. "open source for the people" just does not apply.
Those are mostly for embedded devices and the current "sponsor" is Apple.