Ann: Annual General Meeting Addresses and Presentation, page-59

Dr Lewis stated that "TENNs is a unique BrainChip innovation that replaces the wildly popular transformer architecture."

20 years ago machine learning researchers and Phd students were so into SVM, but now the machine learning computation has been dominated by Transformer, and SVM has been pretty much archived. Thanks to the rise of GPU. Without GPU, there will be no transformer. That is why you can see Jenson Huang invited all inventors of transformer joining his recent aunch event of Nvidia's new AI Chips.

But transformer under the current architiecture based on von neumann theory will not fit for the future of AI computation. The end to end method in a large model based on moving trillions of data between GPU and memory over countless cycles for generating a result, will cost too much energy. Most of you probably didn't know 90% energy used by GPU is for data movement and only less than 10% is used for actual computation. Even if you connect millions of GPU boards together, the computation ability hardly increases.

Therefore, a new computation method is required for future AI technoloty. The big players Sean mentioned in the AGM of course can see this. too. As Dr Lewis concluded "BrainChip is positioned to be at the centre of the next revolution in AI."

I am holding my Brainchip investment to see how the next AI revolution turn out.