Hi SG,![]()
Edge AI Box
Our Edge AI computing solutions help accelerate the deployment of new applications across IoT use cases: security and surveillance, retail, smart factories, smart cities and more.www.qualcomm.com
I was just looking at Qualcomm‘s Edge Box and they have a range of five which makes it interesting from a competitive point of view for Brainchip. Obviously we are targeting the ”Edge” edge but I am guessing we will still be connected to the internet to be able to communicate to the outside world. Eg an alarm is activated and a message goes to dedicated response line/person? So I also believe we will be connected to power to the low power advantage Akida has will be neutralised. I could be wrong of course as I have no technical knowledge on these products.
On the same Qualcomm page I saw GMAC and CVEDIA as partners who we are with also.
Will CVEDIA’s be partnered in this regarding CCTV footage?
It’ll be interesting to see the competitive specs to show the market what Akida is capable of.
I know from a personal perspective I often buy the known larger brand product (e.g. Samsung vs Hisense) as I’m also buying; reputation, confidence in the known brand and after sales assistance; so it will be interesting to see how Brainchip can get a toe in the market against QUALCOMM and NVIDIA etc.
I think price will play a big part in getting the sales but I think we will have to sell for less profit to become known and trusted and build a reputation and brand. That might not be a big issue as we know AKIDA is a lot cheaper than NVIDIA so even at a cheaper sales price the profit margins might not be too bad.
Looking forward to seeing what CES2024 brings.
Good point.
The VVDN EB will include Akida 1 SoC with a WiFi modem and Ethernet interface. I don't know whether the inbuilt ARM Cortex processor will have sufficient processing power or whether it will need an additional processor. Akida 1 has a number of built-in communication interfaces, so the EB could communicate with a local computer or it could communicate over the internet.
Hexagon and presumably its predecessors employ a mishmash of DSP/GPU/CPU for its "AI" capabilities such as ML. This will involve running software. I don't know how much of the classificaftion/inference will run on software, but it will use significantly more power than the VVDN Akida EB.
As you say, that may not be a deal-breaker in powered locations. But running GPU necessitates mechanical fan cooling, adding to both component cost and assembly time, as well as increasing the overall box size.
The VVDN Akida EB will be "greener" that Qualcomm's, using less power and less material.