AGI is 100% possible, even if the current breed of transformer-based models are not it, and even if silicon is not it. There's nothing special about human brains that we won't eventually be able to match (and then exceed) in vitro. We are living proof that intelligence can be built out of matter, and that human-scale intelligence can run on 20 watts. It's not a matter of if, but when.

There is (eventually) no more profit to be made on energy when energy becomes virtually limitless.

There is (still) a lot of profit to be made on half-baked semi-AGI prospects.

That's ok, that's when you change the definition of AGI and claim success!

There is not even an agreed upon definition for intelligence or for AGI.