i think so too, for example i think, that Laptops but also stationary office pc will have a standard AI chip inside very soon, also every traffic light in the world, every surveillance camera that more than just film or take a picture, every machine that transports people, not just cars, even sport devices like ski-bindings and bycicles, climbing ropes, whatever .
Some examples:
1. for classic edge devices its like laptops its obvious, the power-savings alone justify the extra 25 bucks.
2.1 for stationary computers, there are plenty of applications that have a latency problem because they depend on license servers that are too slow and make work annoying (maybe neuromorphic chips could bring new security systems that are faster?)
2.3 further there are a lot of applications that start to become smart but are actually not very smart yet and could benefit from on device ai. When you do technical drawings or creating 2D or 3D models for example (e.g. geometric for plans that go to production or construction site, arcitectural, for calculaton purposes, for quality managment and coordination between planners, visualisations, etc.) you have the problem that all different specialists use different types of information (e.g. points/lines/areas/volumes that are linked with each other). The programs everyone works with have been in development for decades and are obviously becoming better/smarter but honestly i think they are still pretty „unintelligent“ and some more traditional programs are useful but flat out stupid. When I think about the amount of hours (can be counted in days/year/employee) that only the rather small company I work at has to spend time on searching miniscule errors in geoetry that create errors in calculations and plans i really hope that having a slightly intelligent chip could help, because obviously decades of work by the programmers have only shifted the problem around.
3. Traffic monitoring and control: every major and modern city (but also small ones start to have those) have massive traffic control systems. The main problem is that often roads and parking spots are not where they should ideally be and it cant be changed. So what is done instead: Every Information available (parking spot available/not available, every traffic light, every traffic camera, taxis, busses, trams, trains, subways, events that generate traffic, every car that drives in and out of the city and in between regions within the city, etc.) gets processed with the ultimate goal of reducing traffic jams and travel times. Therefore all those devices are interconnected, need sensors, and most of those things are very old fashioned… not programming but the hardware: with cables and wires in the ground that measure magnetic fields to verify if a car/truck/bycile/pedestrian just passed because image recognition wasnt availabe when those systems were built and even now are not backed that way because the chaos and danger that would follow a is just too large… People trust the green light. Imagine the cost savings, for example check out what it costs to have an electrician install a couple of meters of cable, you‘d be surprised…
3. i could go on for hours… about other topics
i‘m not saying akida everywhere, just that the demand for low to ultralow power SoC AI chips will be truly enormous… and I think all estimates of the size of this market could be way too low. So even if it‘s less that 1% akida it will still be enough.
good luck and best regards
Expand
(20min delay)