According to PwC, AI could impact global GDP by $15 trillion! But this scale is only possible due to trillions of IoT edge devices. Nandan Nayampally, CMO at BrainChip, joins Ryan Chacon on the IoT For All Podcast to discuss the evolution of AI in IoT or AIoT. They cover the use cases that have emerged with edge AI maturity, the difference between an AIoT solution and an IoT solution, the markets that will accelerate AIoT adoption, and advice for how to prioritize a market to focus on.

About Nandan

Nandan Nayampally is an entrepreneurial executive with more than 25 years of success in building or growing technology businesses with industry-wide impact. He was at Arm for more than 15 years in a variety of product marketing and product management leadership roles, eventually becoming vice president and general manager of Arm’s signature CPU group and the Client Line of Business where he identified key technology investments, developed a strategy and roadmap of products to deliver compelling, market-leading solutions for billions of SoCs, while establishing strategic partnerships and alliances. Nayampally comes to BrainChip from Amazon, where he helped accelerate the adoption of Alexa Voice and other multimodal services into third-party devices.

Interested in connecting with Nandan? Reach out on LinkedIn!

About BrainChip

BrainChip is the worldwide leader in edge AI on-chip processing and learning. The company’s first-to-market neuromorphic processor, AkidaTM, mimics the human brain to analyze only essential sensor inputs at the point of acquisition, processing data with unparalleled efficiency, precision, and economy of energy. Keeping machine learning local to the chip, independent of the cloud, also dramatically reduces latency while improving privacy and data security. In enabling effective edge compute to be universally deployable across real world applications such as connected cars, consumer electronics, and industrial IoT, BrainChip is proving that on-chip AI, close to the sensor, is the future, for its customers’ products, as well as the planet.

Key Questions and Topics from this Episode:

(00:46) Introduction to Nandan and BrainChip

(03:54) What new use cases have emerged with edge AI maturity?

(06:12) The evolution of AI in IoT (AIoT)

(08:18) AIoT solution vs IoT solution

(11:02) What markets will accelerate AIoT adoption?

(14:05) How to prioritize a market to focus on

(17:48) Learn more and follow up


– [Ryan] Hello everyone and welcome to another episode of the IoT For All Podcast. I’m Ryan Chacon, and on today’s episode, we’re going to talk about the evolution of AIoT, key market expectations when it comes to what is accelerating adoption, and what does AIoT solution look like versus an IoT solution.

With me today is Nandan Nayampally. He is the CMO of BrainChip. They are a worldwide leader in Edge AI on chip processing and learning. Fantastic conversation. I think you’ll get a lot of value out of this one. Before we get into it, hit the bell icon, so you get the latest episodes as soon as they are out, subscribe to our channel if you have not done so already, and give this video a thumbs up, and we really appreciate it.

Other than that, let’s get onto the episode. Welcome Nandan to the IoT For All Podcast. Thanks for being here this week.

– [Nandan] Thanks, Ryan. It’s a pleasure to be here.

– [Ryan] Yeah, it’s great to have you. Let’s kick this off by having you give a quick introduction about yourself to our audience.

– [Nandan] Hi so I’m Nandan Nayampally. I’m the Chief Marketing Officer and Head of Product for BrainChip. I’ve had over a 25 year career, nearly 30 year career in the semiconductor space initially with AMD. A long time with Arm spend some time with Amazon on the Alexa team as well. That’s in the nutshell what I do.

And if you’d I’ll give a little bit of a, an intro to BrainChip itself.

– [Ryan] Yeah, that’d be great.

– [Nandan] So BrainChip is a company that’s been around 10 years. Started by a a couple of a few scientists and engineers who decided to go after modeling the brain function, primarily because the brain is probably the most efficient computer known to man. The entire brain effectively takes about 20 watts and does way more calculations than we can actually imagine.

And it’s actually the fundamental thing about how we get towards artificial intelligence. And of course the everybody’s key goal at some point to get to AGI, right? Which is artificial general intelligence, artificial general intelligence. That’s the mission behind what BrainChip started.

But they also focused on bringing AI to the edge, and so we built processors with the neuromorphic background, which means they’re built more like the brain function, more like neurons rather than just a bunch of parallel wide vector computers. And the whole intent is to actually only compute when necessary, only communicate when necessary and be extremely efficient.

And that’s actually the sense of why you need to do those things at the edge. Today, majority, almost all of the real computation for AI happens on the Cloud. And that’s a challenge because it doesn’t always scale. As you get more complex things happening at the edge, you want it to be closer. You need to be real time.

Even with 5G and promise of almost infinite connectivity and bandwidth, it really is not fast enough. Additionally, there are things that you want to keep close to the device. You want it to be private. You don’t want all your data going out. And anytime you think about more and more complex things happening closer to you, closer to the device you really want them to be computed close.

For that, we need it to be efficient. For that, you need to be cost effective. And that’s the real motivation for what BrainChip’s trying to do.

– [Ryan] Have you seen because what it sounds like from my conversations that I’ve had when it comes to the growth of edge computing and IoT, being able to add in the AI, the ML elements at the edge, not only like you mentioned, things can be done faster. You can usually, it’s usually cheaper. You can scale more easily.

Have you seen the growth of any, new use cases or applications that maybe required that? Maybe but maybe were too far ahead of its time in the sense that they use the technology wasn’t there yet but now it is because of how far edge computing has come. And if so, where have you seen that growth?

– [Nandan] So that’s, that’s an excellent question. You’re absolutely right in that initially there wasn’t enough compute power at the edge to do kind the kinds of things that needed to be done in time. So it is initially first driven a lot more by need. And for that I would take cars as an example, right? ADAS or drivers assist autonomous drivers assistance.

As you started getting closer and closer to things that need to be done in milliseconds or microseconds because of the timebound nature, you started doing that. The advantage that you had in cars are pretty big things. So you do have enough power and compute power that you can put into it to handle those.

So that’s how it started. Then you started seeing things in the drone space. You started to see, you’re seeing things in computer vision. You started seeing things in fleet management. Now you’re starting to see things in healthcare. Even in healthcare, you’re seeing sensors still sending more to cloud, but now you can start doing a lot more things closer to the device and that’s is beginning to open up the market.

– [Ryan] The capabilities and what’s possible. Absolutely. So one of the new, like a term that’s been thrown around for a little while, but starting to become more popular now with the growth of AI is just AIoT we’ve been talking about, and obviously we focus a lot of our efforts on IoT, we’re getting more into AI because of its popularity, its role in benefiting from IoT data on the enterprise side particularly.

Talk to me about the evolution of AI and kind of where what, what what’s changed or what’s evolved there to make it more of a thing now and then what markets are you really expecting to see accelerate the adoption when it comes to AIoT?

– [Nandan] That’s a fair question. So if you, I mean you, you’ve been an IoT person for a while, right? So IoT started with when you had connectivity with some sensing, and then you added compute to it. So those three things made an AIoT, no I mean an IoT node. Now the compute starts becoming artificially intelligent rather than just procedural.

And so if you think about the impact of AI as a whole, PricewaterhouseCoopers released a study saying their expectation that it would affect the global GDP annually by 2030 by about $15 trillion. Now, that’s not going to happen in the Cloud for that because even if you think take all the data centers in the world, you’re probably gonna have, by that time, 12, 15,000 data centers.

It’s really, it’s going to come from the trillion plus edge devices. So really it’s the artificial intelligence of things because the only way to scale AI is not to actually centralize it but to decentralize it and distribute it. And so there are numbers that say the AIoT market in 2030 could be anywhere between 1.2 trillion to 2.5 trillion.

And that’s a lot of that is going to come from very differentiated, intelligent devices at the edge. In services in the cloud that complement that. It’s not one or the other. It’s both, but really actually the proliferation of the edge devices improves the global scale of AI.

– [Ryan] Gotcha. Okay, fantastic. And let me ask you this. When we talk, or I guess in our past conversations here, we talk a lot about IoT solutions, how they work, how to do them well, increase your likelihood of success, how the technologies are playing a role in different industries, you name it. And now when we get into AIoT solutions, how do, when you’re thinking about that, or talking to somebody about an IoT solution versus an AIoT solution, what is the difference?

Is it really about just the AI models being worked in. Being able to benefit that solution at the edge or in different, different pieces of the solution or how do you think about an AIoT solution versus an IoT solution? What’s the main difference? How should people be thinking about that when they hear it?

– [Nandan] Yeah, so I think you hit a couple of those points on the head, right? AI, IoT, the main thing for it was you had enough compute, perhaps something around the sensing, but really it was the connectivity to the internet that made it AIoT, I’m sorry, IoT, right? With AI you need to be able, for most part, to do what needs to be done on the device itself.

The internet connection’s not the main part of it, right? The internet connection after that actually helps to scale it globally and you may not, the main thing is you’re not thinking about sending raw data to the Cloud to get it computed. You’re doing most of what you need to do on the device. That way you have security, you have privacy because you only send out data that you need to, and it’s probably metadata rather than raw data and certainly not sensitive data.

So when I think about AIoT, it’s really about doing intelligence real time, very close to the device, being secure, being private. Minimizing the amount of traffic that it generates, right? And at the same time, giving both the user of the device a very new experience a real experience, but also scaling to cloud to have a broader AI solution.

– [Ryan] Yeah, absolutely. Yeah it’s interesting just to seen over the last year, we’ve obviously seen tons of popularity when it comes to AI on the consumer side, and AI has always been around with, in enterprises and stuff, but it’s becoming more prominent. I think a lot of, because people are paying more attention due to what happened last year into this year.

We on our side just seeing the growth of popularity in AI content, especially on the enterprise side. So it’s fascinating to see. And it’s nice to be able to chat about how AI and IoT are working together and how it will work together to benefit basically everyone involved. You’ve already touched on this a little bit, but do you see any markets that maybe a normal, a person wouldn’t necessarily think about that are going to really lead the way in adoption when it comes to AIoT solutions?

– [Nandan] Yeah, I think this is not a far reach because you’d expect this, vital science prediction, chronic disease prevention, that kind of or at least preventative maintenance for health actually becomes a very interesting point with AIoT, especially if you can manage the security, privacy concerns people have.

And now if you have enough compute power on the edge device, rather than having to send the raw data to cloud to compute or to your cell phone or to your local hub. If you can do that, then you, then it frees you up to do what you need to do while having a reasonable level of confidence that you are in good hands, so to speak.

Healthcare is clearly one. I see agriculture, which is again not surprising, but agriculture is where you can actually utilize localized AI algorithms, whether it is for, weed maintenance and whether it is for recognizing that there’s potential frost impact. There’s so many things that you can do.

15 years ago there was a concept called smart dust where there were these processors that were like dust, right? Really small but connected. Now you can have much more intelligence and autonomy in some of those processors to help you not just sense, but probably dispense what needs to be done to protect it. Infrastructure for preventative maintenance of factories, clearly, big return on investment.

And so there are so many different areas cropping up. Obviously security surveillance always, you know, talked about. You can do more of it. And there’s a dime a dozen new areas cropping up. Smart cities, smart homes that are really beginning to utilize the fact that not only can you do intelligent inferencing as you say, Hey, have a model that works.

I’m going to inference and return. But with neuromorphic capabilities, which is what BrainChip, for example, does, you can also learn and customize on the edge without having to go to cloud to retrain. This training is an expensive thing for models to do. For example, GPT-3 took $6 million, I think in four weeks.

Anytime you want to change something you’re almost cost prohibitive to try retraining it. So the more you can do in terms of customization and extension on device, the better it is for the overall service as well.

– [Ryan] Yeah, makes total sense. You mentioned a lot of different industries in different areas and how. I wanted to it’s slight little pivot here, but I wanted to ask you how companies are listening to this. They might obviously not all be in the same world as as your company is, but they all have to hone their focus on different industries to devote their time, resources, and such to, to build their business around.

How have you found it or what advice would you have for companies on how to prioritize the different markets to focus on? Because obviously that goes into who you build for, how you build, what you build. And there’s a lot of excitement that we’ve been talking about in different industries for this conversation and many others.

But how would you, if I’m listening to this, how would you assess what market is best to focus on? Like how, what kind of advice would you have for companies to, to do that or how to approach it?

– [Nandan] So sorry. Let me ask you a clarifying question, right? So by that do you mean hey, you are a semiconductor vendor trying to sell in to this market? Or are you saying any company that is trying to build for AIoT? So let me get a bit into what you’re trying to do.

– [Ryan] I would say more the latter. So if I’m a company that is trying to build a component for an IoT solution, an end-to-end solution, has AI technology that we want to build out for a particular industry or use case, but maybe, when they’re starting out, they realize this is applicable to lots of things.

How do you advise them to stay focused or pick where to start?

– [Nandan] So usually when you’re bringing a solution to market, right? The first thing you look at is, am I making a meaningful difference to what it was before? Two ways to do it. Either. Am I bringing a capability that did not exist, right? Or am I making something so much more cost effective that it can scale bigger?

There are others as well, right? But I look at products in two basic ways. And so it’s, if it’s the former case, right? That I’m bringing something to market, obviously you have defined a real pain point where the market, it’s been asking for a real change. My word for this is, are you bringing an aspirin or is it a vitamin?

Are you just making something feel better or are you really solving a headache? So focus on the ones that truly solve a headache. And if you do that, then naturally you will get faster traction. And once you get proof points, then you can actually scale it broader. But you need to have clear wins that showcase how it’s done.

That also helps you build out an ecosystem and credibility, and then you can scale from there, right? And so focus on a core competencies. Partner with the guys that are really trying to break new boundaries. And then once the tip of the spears goes through, then follow through.

– [Ryan] Yeah, that’s great advice. It aligns with bits and pieces of other conversations we’ve had in the past, and I think, you really summed it up in a great way for our audience to understand what they, what and how to focus their initial efforts on when it comes to building and bringing something to market.

Appreciate you touching on that. This is, this has been a great conversation. I really appreciate you taking the time being able to dive into the AI, AIoT side of things. Talking about it from, a higher level and explaining how it’s really working is great value. So I really appreciate you taking the time and it’s, it sounds very exciting, all the things you have going on on your end as well.

For our audience who wants to learn more about what you all have happening over there, follow up with any questions, reach out or anything like that, what’s the best way to do it?

– [Nandan] Just go to or That’s quite straightforward. It’s spelled exactly like it’s said, right? Brain chip one word dot com. And contact us from there. We look forward to inputs, questions, feedback. And thanks for giving us a chance, Ryan. This is a good forum.

I really appreciate I, I’ve had fun doing this interview as well. Conversation’s good. I look forward to your viewers actually reaching out and connecting with us.

– [Ryan] Sounds good. Well really appreciate your time and thanks again for being here.

– [Nandan] Thanks.

Hosted By
IoT For All
IoT For All
IoT For All is creating resources to enable companies of all sizes to leverage IoT. From technical deep-dives, to IoT ecosystem overviews, to evergreen resources, IoT For All is the best place to keep up with what's going on in IoT.
IoT For All is creating resources to enable companies of all sizes to leverage IoT. From technical deep-dives, to IoT ecosystem overviews, to evergreen resources, IoT For All is the best place to keep up with what's going on in IoT.