the important part of the puzzle is why it is taking so much...

the important part of the puzzle is why it is taking so much time for someone to sign a new contract with brn and to that effect my understanding is memory of our chips.
we have very limited memory, SNNs and TeNNs work as in memory commute and for that model size have to extra small. That is why whenever we see a demo with SNNs or TENNs model size becomes 300 times small, parameters become even less.
The reason for Gpt success is one model fits all but the model size is too too big.
So that is why everyone is trying to get small language models where model size can be reduced to specific requirements and energy burn, latency can be minimised. Brainchip also start working on that end and looking at the track record of brainchip it seems like we should be able to crack a few things by end of 2024, and once we can provide efficient model size to specific industry we will start getting deals.
So yes there are a few more slips between the cup and lips but hopefully we will be able to enjoy our coffee.
dyor