Don't count your chickens before they hatch. I believe that the odds of an architecture substantially better than autoregressive causal GPTs coming out of the woodwork within the next year is quite high.
How does that equate to "winner take all", though? It is quite apparent that as soon as one place figures out some kind of advantage, everyone else follows suit almost immediately.
It's not the 1800s anymore. You cannot hide behind poor communication.
It's not the 1800s anymore. You cannot hide behind poor communication.