Hi Frangipani,
So, it's an indisputable fact that BrainChip's AKIDA is unique in that it is the only commercially available neuromorphic processor with on-chip learning, since neither Synsense nor Innatera offer on-chip learning. In that sense, Manny was on the right track IMO.
Regards,
Bravo
In that sense?!
It is an indisputable fact that the above is not what manny100 had actually said.
Just to refresh everyone’s memory:
Yes, we BRN are unique as we offer the only 'cloudless' commercial solution ATM - this is an indisputable fact
Speaking of on-chip learning: Loihi 2 is fully digital and is claimed to have on-chip learning - like or similar to Akida.
Yes, true - Loihi and Loihi 2 have been described as research chips by Intel and are not yet commercially available as physical chips to date. While Intel’s next-gen silicon may still be a couple of years away from what we’ve gathered through interviews with Mike Davies, I’ve been wondering whether Intel might actually consider offering the licensing of Loihi IP much sooner, depending on customer demand. After all, the idea of IP offerings for Intel Foundry customers has been around for quite a while - see this slide taken from a September 2021 presentation, when Loihi 2 was revealed:
As Hawaii’s Kilauea Volcano erupts, Intel is announcing its Hawaiian themed LAVA framework and its Loihi 2 neuromorphic chip on Intel 4
www.servethehome.com
In a similar presentation six months later (March 2022), the word
potential had been deleted from the presentation, and the offering had been rephrased as “IP and embedded accelerators for Intel Foundry customers.”
This brings me back to our engagement with Mercedes:
While we BRN shareholders naturally all wish for Mercedes to announce BrainChip as their future neuromorphic partner, we should keep in mind that Markus Schäfer praised both Intel and BrainChip in his post on neuromorphic computing last year, and that employees from both companies liked his post, which could mean all of them had high hopes at the time.
This observation in combination with the word
potential in “positioning for a potential project collaboration with Mercedes” showing up in a former BrainChip intern’s CV, suggested to me that Mercedes must have been weighing their options and were evaluating more than one neuromorphic processor last year.
That is why I wrote on January 29:
As for Loihi not being commercial, yet, see my comments about the potential licensing of Loihi IP further up - I hadn’t yet spotted those presentation slides when I wrote my earlier post.
Also, I’ve never understood why Mercedes would make such a big secret out of continuing their collaboration with BrainChip, after they themselves had already let it slip in early 2022 that their AI engineers had implemented Akida for keyword spotting in the EQXX and hence had - intentionally or unintentionally - lifted the NDA veil. The whole world has since known about their engagement with BrainChip anyway, so what was the point of draping those blackout curtains over that NDA again?
The Mercedes top management has since named several start-ups (unrelated to neuromorphic tech) as their partners in LinkedIn posts. So why the ongoing secrecy regarding the carmaker’s future neuromorphic technology partner? Unless they are still weighing their options? This would not even contradict the recent statement by Sean Hehir “We have not lost to any competitor
at this point in time. We’ve seen some small delays, as people need more time to evaluate…” (from 12:12 min in the Quarterly Investor Podcast, Episode 5 - Q1 2024)
And now there is also Innatera as yet another competitor - I happened to check their website for partnerships and discovered this:
Not sure how they define being an “industry partner”, but if those were exclusively involved in super secret NDAs with the Dutch start-up (founded in 2018), wouldn’t they mind having their logos prominently displayed on the Innatera website? Oh, and plenty of other familiar names showing up here, as you can see: Ericsson, Infineon, Schneider Electric, BMW, Safran…
Those “partnerships” may not all lead up to a real deal getting signed eventually, but they are evidence that the competition between the various neuromorphic players is heating up, despite our first-to-market advantage. And we all know that it is not always the best technology that wins the contract.
But is that a reason for BRN shareholders to be pessimistic? Not at all.
After all, the future of BrainChip is not limited to a binary option of either world domination or insolvency. For various reasons, it is just taking a lot longer than most of us anticipated.
I continue to trust in the optimism reflected in our company’s reply to manny100’s question:
"We know the market is heading in our direction, and the ability to process data on The Edge is something that more and more AI applications will be expected to perform. In that case, we are uniquely positioned to capitalise on that trend."