Sounds like Kevin is being very careful not to imply certain things on IBM's behalf, which is understandable. If he says “only Akida” it shifts the conversation from architecture to vendor endorsement, which he seems to be deliberately avoiding.
Having said that, he’s very explicit and forthcoming about Akida’s performance and power numbers, and he repeatedly names Akida as the tool that proved right for this workload.
What strikes me most is what Kevin has to say about IBM's plans in terms of potential productisation. When he says “On the IBM question, I can't speak to IBM's product plans. The market drives that conversation". I think he's making it pretty clear that IBM won’t commit to product plans until customers start demanding it.
And that’s where it seems like a bit of a stalemate type situation to me. "Stalemate" is probably not the right word I'm looking for, rather that it highlights a real gap that now needs to be bridged.
BrainChip presumably can’t rely on a Principle Technical Specialist from IBM to sell its technology. Equally, IBM isn’t going to productise or promote it unless customers are already asking for it. And customers generally won’t ask for something they don’t clearly understand, can’t easily replicate, or haven’t seen deployed commercially elsewhere.
So, maybe the next step is for BrainChip to develop this into something repeatable and commercially legible under its own steam, by
documenting a clear reference architecture that others can reproduce. Then start demoing it and promoting it, so customers know it exists, which in time might lead to paying customers and eventual revenue.
In my view, Kevin has already done BrainChip a significant favour by validating a real, enterprise-relevant use case inside a credible production environment, backed by impressive numbers.
From here, the outcome may depend far more on BrainChip’s follow-through than on IBM’s intent.
brainchip.com
thequantuminsider.com
