The IoT Show · 2025-12-12 · 17 min
Key moments - from our scoring
Substance score
34 / 100
Five dimensions, 20 points each
Leonard positions the IoT industry at an inflection point where embedded technologies and AI are creating viable commercial applications after years of theoretical promise. He distinguishes between "IoT for AI" - using IoT data to train AI systems - and "AI for IoT," where machine learning deploys closer to sensors and endpoints with increasingly capable microcontrollers. The conversation emphasizes contextual computing and digital twins that aggregate sensor data into rich, machine-readable contexts rather than simple data points. As semiconductor companies like ST Micro, NXP, Analog Devices, and Qualcomm consolidate IoT, AI, and edge computing stacks, Leonard argues the focus should be on developer experience and viable end-market value rather than religious adherence to openness. Marc contributes industry observations including that modern embedded stacks should unlock across hardware platforms and that the best automation requires no dashboard. The discussion targets practitioners building edge AI and IoT systems who need clarity on which technical and business consolidation strategies actually deliver scaling value.
IoT for AI involves using IoT-collected sensor data to train and contextualize AI systems, while AI for IoT means deploying machine learning capabilities directly on edge devices and microcontrollers closer to the data source.
Contextual computing allows sensor data to be aligned to machine-readable ontologies that provide rich environmental understanding to humans and systems, replacing simple isolated data points with integrated information that drives better decision-making.
Semiconductor and software companies are consolidating IoT, AI, and hardware stacks to simplify technology integration, reduce costs, and create viable commercial applications - prioritizing developer experience and end-market value over strict adherence to open standards.
Technical capability alone was insufficient; many IoT projects failed because they lacked economic viability and clear paths to end-market value, a challenge Leonard believes is now being solved through better integration and simpler developer experiences.
IoT technologies are now hitting inflection points where AI, edge compute, and better developer tools converge, allowing embedded applications that were theoretically possible but economically unviable in the past to now generate real business value.
Our reviewer’s read on each dimension, with quotes from the episode.
The episode contains scattered conceptual framings (IoT vs. AI for IoT, contextual computing, trough of disillusionment) but largely retreats into vague abstractions and motivational puffery. Most substantive claims lack concrete examples, timelines, or measurable outcomes. The second half devolves into random aphorisms and bar-talk philosophy with minimal actionable insight for an operator.
There's IoT for AI and then there's AI for IoT. Um, yeah, I mean, that sounds kind of cliche, but that is sort of the dynamic that's playing out, right?
I really do think that IOT is going to get a second win
The framing of 'AI for IoT' versus 'IoT for AI' is presented as novel but amounts to well-worn industry observation. References to Gartner's trough of disillusionment are dated playbook material. The latter portion relies on bumper-sticker philosophies ('best dashboard is no dashboard') that lack substantive differentiation or first-principles argument.
There's IoT for AI and then there's AI for IoT. Um, yeah, I mean, that sounds kind of cliche, but that is sort of the dynamic that's playing out, right?
I think we've been kind of in the trough of disillusionment with Iot
Leonard appears to be an analyst at Next Curve with industry credibility and runs IoT Coffee Talk. However, his actual operational track record at scale is never established in the transcript. Mark's sudden appearance and affiliation (recently joined Qualcomm) is mentioned but he contributes minimally and late. Neither guest demonstrates hands-on execution experience with the problems they discuss.
Leonard is an analyst Next Curve, uh, is his company.
you uh, know you, you joined Qualcomm recently. Qualcomm.
The episode names semiconductor companies (ST Micro, Analog Devices, Qualcomm, NXP, Nvidia, AMD, Intel) and industry events (Embedded World, IoT Stars) but provides zero metrics, case studies, customer examples, revenue figures, or deployment timelines. Assertions about consolidation and 'inflection points' float unanchored to data. High abstraction, minimal specificity.
whether it's ST Micro or it's. I, uh, mean even Analog Devices, Qualcomm, nxp, all these guys now trying to, to bring this stat together
There's no such thing as the shell hardware. Everything ends up custom
The host asks open-ended questions but rarely probes deeper or challenges assertions. Follow-ups are surface-level ('tell me how things changing') with minimal push-back. The conversation meanders into bar banter and random quotes at the end without coherent thread. No productive disagreement or Socratic challenge to sharpen thinking.
So tell me. Let's narrow the conversation down, Iot. Ar, okay, in a concrete way, how things changing and how fast are they changing
No, no, not yet. Not yet.
Computed from the transcript - who did the talking, and the words that came up most.
When you start a conversation with Leonard Lee (neXt Curve), you never really know where it will go, but what's sure is that it will be interesting!This one I had with him during the last IoT Stars networking event didn't disappoint and I could ask him about the latest trends he is seeing in IoT, Edge AI, and more. And towards the end of the interview, we had the chance to have Marc Pous join in! Resources:Stay up to date on latest tech trends on Leonard's channels:
Transcribed and scored by The B2B Podcast Index.
Speaker A: You know, the party is on when Leonard is in the house. And Leonard was definitely in the house at IoT stars when we were doing this live streaming at Embedded World in Anaheim recently. Leonard is an analyst Next Curve, uh, is his company. You will find a lot of information on his channels, LinkedIn and others. He also runs the IoT Coffee Talk with Rob Tiffany that I attend, uh, to sometimes. Really interesting conversations around IoT. I really recommend you check it out as well. Uh, and, uh, well, Leonard has lots to say and it's always fascinating to have a conversation and start him, um, on some ideas. Edgy. How are things going there and what's the trend there? I really enjoyed a conversation with him. Hope you'll enjoy it as well here on the IoT show. If you like the episode, do not forget to subscribe to the channel. Give us a, like, put a little comment down there. And once Again, thank you, IoT Stars, for supporting the IoT Show. IoT Stars is the IoT networking event you want to attend. Uh, if you are remotely interested in IoT or have an IoT project and want to meet the right person, have the right conversation. Conversation that happens at Iot, uh, stars events.
Speaker B: Oh, my God, I look terrible.
Speaker C: That's life, man.
Speaker B: I know.
Speaker C: We look good, man. We look good.
Speaker B: Oh, um, my God.
Speaker C: Well, we're life. We have few people online. Thanks for joining in.
Speaker B: Hey, how's it going, everyone?
Speaker C: Leonard is here. The Leonard, as usual with Leonard. We don't know what we're going to talk about.
Speaker B: I have no idea.
Speaker C: But that's gonna be fun. We, ah, are at, uh, Embedded World. Yeah, this is the IoT star networking event.
Speaker B: Yes.
Speaker C: Great. People waiting for us, you know, holding our beers down there.
Speaker D: Yes.
Speaker B: Um, a lot of traffic. Yeah, I just got up here from San, um, Diego. Yeah, I was at a Cisco event and so it was all about networking and security and all this AI supercomputing nonsense. And then, uh, now I'm here. Yeah, it's all about embedded in IoT, the stuff that matters, actually. What? I think it's going to become super sexy.
Speaker C: Like, jokes aside, don't you feel like.
Speaker B: Wait, wait a minute. What makes you think I was joking? Do I look like that's actually joking?
Speaker C: Uh, let me actually rephrase. This was not a joke. Here we are having real conversation about real stuff, right? Oh, yeah. More so than in other networking events. What other.
Speaker B: Oh, yeah.
Speaker C: Say politely. Right?
Speaker B: Yeah.
Speaker C: So what are the things that you're. You're observing the trends of the industry. You're Actually advising companies, uh, helping them understand what's next. Right. Next Curve is the name of the company.
Speaker B: Wow. Wow.
Speaker C: Am I good?
Speaker B: Yeah, you're. You're making me look a lot smarter than I really am.
Speaker C: No, you are, you are. The idea is to be smarter than the average of the group.
Speaker B: Yeah, well, yeah, I mean, that's, uh. If you call your company Next Curve, yeah. You're kind of obligating yourself.
Speaker C: So tell me. Let's narrow the conversation down, Iot. Ar, okay, in a concrete way, how things changing and how fast are they changing when it comes to that combination of these two technologies.
Speaker B: Now, I think it's important there's like two things. There's IoT for AI and then there's AI for IoT.
Speaker C: Okay.
Speaker B: Um, yeah, I mean, that sounds kind of cliche, but that is sort of the dynamic that's playing out, right? Because what we're seeing is a lot of this wonderful AI technology, largely, uh, machine learning that's benefiting from much more capable compute that's coming down literally to the microcontroller. And so you're able to deploy much more capable, quote, unquote, AI closer to the sensor, closer to, uh, the endpoint device. And so, um, that, I think that's a, uh, that's really, um, catalyzing dynamic, or at least one aspect of a catalytic dynamic that we're seeing right now. And the other one is IoT for AI. And so one of the things that I think is going to be pretty huge, um, very soon, in fact, I would say that it's happening right now is contextual computing. Contextual. And, you know, some people call it contextual AI, but it really is more like contextual computing. How do we, uh, enable AI to contextualize what is being sensed and gathered from environments, right. Business environments or whatever kind of environment, um, that are being captured by sensors and then all the connectivities, ah, architectures and technologies that allow you to aggregate all that stuff, stuff, you know, um, align it to an ontology that makes it machine readable. Once it's machine readable, of course, you can, you know, go through a contextualization layer that allows humans to understand in a much more rich way what's going on. And that's, that's really what like, you know, Rob Tiffany always talks about it. It's about the clipboard, right?
Speaker D: Yes.
Speaker B: You know, you get rid of the clipboard, and it's not just about little data points, simple stuff. Now you're able to bring together a much richer collection of information together from the edge. Uh, and, um, do some really funky stuff.
Speaker C: Yeah. Things that we thought possible but didn't know how to do. Now we have different technologies and bricks coming together. Digital twins that allow you to model and contextualize IoT that's feeding environmental data, AI that works, analyzes all that.
Speaker B: Yeah.
Speaker C: We just talked about software, defined connectivity, something else here. All these things are coming together.
Speaker D: Yeah.
Speaker C: Kind of magic.
Speaker B: Yeah, yeah. Uh, yeah, it's great. I, uh, mean, you know, I think I want to give kudos to a lot of engineers out there. I love you guys. Right. Mad respect for all of you. Um, I really appreciate everything that you guys do and a lot of, like, the challenges you face working with these technologies and trying to express them in terms of, you know, the business value that all these ridiculous, uh, you know, forecasts of exponential tams and all that stuff, uh, suggested. But the thing is, you guys knew. The thing is, now technologies are hitting inflection points that allow what you thought would want it to happen to happen. Right. And so there's this new value, opportunities that are availing themselves. And I think, um. What is it? We've been kind of in the trough of disillusionment with Iot. I mean, I really do think that IOT is going to get a second win.
Speaker D: Right.
Speaker B: And so embedded technologies again are going to have now a different role. When you look at it in terms of like a cyber physical system and what you can do with those things and what kind of applications you can enable.
Speaker D: Right.
Speaker B: And, um, I think it's the beer talking now.
Speaker C: You want to shut me up, I think. No, no, not yet. Not yet.
Speaker B: Am I sounding too smart?
Speaker C: I think you're sounding, not only are you Sonics smart, but that makes me want to ask you the follow. Oh, no, Iot stars 2030. Ah. Uh, where are we going to talk to? I'm seeing a trend where this consolidation of different types of stacks out there. So Quantum acquired ed, Impulse, Sand, Arduino, Uh, they're, they're kind of consolidating Mordic Semiconductor M. They acquired Newton just like them, a tiny amount of technology as well. We're starting to see consolidation, tying AI at the edge with IoT and cloud. And so IoT starts 2030. Who's going to be in the room? What are we going to talk about?
Speaker B: Yeah, it's really an interesting question. I mean, I think that's a really good observation because I think consolidation makes sense provided that there is scale that could benefit from it. Right. And, um, at this point, I think we're really looking at an effort to simplify things and seek scale. So yeah, I, I think, um, I don't think there's any. It's inevitable that you're going to have some consolidation, integration in order to squeeze out costs. Right. And to be able to allow uh, applications to be viable. I mean that was like one of the biggest challenges we had in IoT for more than a decade is viability. It's great in terms of like ah, you know, a technical capability, but was it viable? Right. From an economic standpoint. And so yeah, you're gonna, you're gonna probably see a lot of the, um.
Speaker C: I mean you do see a lot
Speaker B: of the semiconductor companies, whether it's ST Micro or it's. I, uh, mean even Analog Devices, Qualcomm, nxp, all these guys now trying to, to bring this stat together. Yeah, Nvidia. Yeah, arguably, um, you know, amd, all the guys. Right, uh, intel, right. Trying to create these, um, let's call them value pipelines. Right. And that involves simplification, it involves integration.
Speaker C: And
Speaker B: a lot of that has to do with kind of bringing IoT together with AI pipelines. And so that's what we're seeing with some of these um, uh, plays right where um.
Speaker C: And they're much needed because like all these technologies are very complex. Integrating them is not simple. Now you'll get options out there the same way cloud simplified things. You can go to AWS and things are done a certain way. You go to Azure, things about a different way. So you basically get options that simplify this aggregation of technologies to implement their solutions. I think Mark is like hovering.
Speaker B: Oh my God. Are you gonna ah, actually be surprised? I was in the door.
Speaker C: Yes.
Speaker D: I didn't see him joining.
Speaker B: Oh my God.
Speaker D: How did you get in here?
Speaker B: Uh, you know, somebody, uh, Brandon recognized me and then he goes, oh yeah, Leonard Lee, you need to talk.
Speaker C: Yeah, we had special instruction though for not letting him in.
Speaker B: Like, yeah, look out for that crazy Asian guy from.
Speaker C: So we were in a very profound conversation about the trends, you know, of IoT and, and consolidation of stacks, IoT, AI hardware and so on. You uh, know you, you joined Qualcom recently. Qualcomm. Yeah, consolidation. But yeah, no, that's, that's fascinating.
Speaker B: Yeah, I mean, but I, I think um, it's something that is inevitable. Uh, it's not something that I think developers really need to have too much angst for because there are going to be, they're going to be stove pipes, they're going to be like competitive stove pipes and so, uh, and then openness. Openness is, it's great, but it has to come, it has to drive valuable interoperability and it's really more of an agenda of interoperability more than open. Right. I mean we've seen this time and time again, uh, don't overdo the open thing. What's most important for a developer community is being able to channel technology applications into end market value. Right. And again, going back to, you know, some of the fixations we had with IOT in the um, first or second round is hey, it has to be open, it has to be, we have to be interoperable in order to scale. Well a lot of those theories didn't play out and so, you know, um, one of the things that I would highly recommend Both the Edge AI community and the IoT community and the combined IoT plus Edge AI community is, you know, uh, don't get too fixated on that stuff, don't get religious about things. Right? We really have to focus on how do we move the community, how do we then support the technology to. I feel like the beer is,
Speaker D: maybe I should drink more beer.
Speaker B: But uh, how do you get, how do you create that dynamic that um, channels that technology into those applications? And scaling is like, you know, we've talked about IOT coffee talk forever. Scaling is a weird concept. You have to be very clear on what you mean by that because scaling in one context is different from another. And uh, the strategies that you come up with related to driving scaling, whether it's for your business or it's for the technology or the service, those are different, they're not the same. Right. And so you can't generalize things like that. So hopefully a lot of these things that we've learned in iot, we um, we um, applied them. The lessons learned in this next round, this next opportunity, I, you know, as a, as um, some encouragement to the community. Yes, there is new opportunity here and there are high value uh, opportunities in you know, some, you know, industries that have a crap ton of money, like the semiconductor industry right now, you know. And so, you know, those of you who follow me know that I do a lot of stuff in the semiconductor industry, manufacturing.
Speaker D: The answer for this question, we have some pretty um, stars. And one says if your aviation technology has a steep learning curve, it's a bad product. So probably it is uh, like the conclusion of the summary of what you said
Speaker C: sounds some good timing. Thanks Mark.
Speaker D: There is no such thing as the shell hardware. Everything ends up custom yeah, no, I agree.
Speaker B: A lot of it becomes bespoke.
Speaker C: Yeah.
Speaker D: And you will love this one. The best dashboard is no dashboard. It's automation that matters.
Speaker C: But some philosopher has been writing this. So now here's the question. Who created this?
Speaker D: It does. It's Lawrence and me.
Speaker B: Oh, my God.
Speaker D: The key to edge AI adoption is better developer experience. What do you think is the key for AI adoption?
Speaker B: Um, yeah, I mean, that's definitely greasing the wheel, but, you know, again, it has to be direction. You have to be, uh, moving in the right direction.
Speaker D: A modern embedded stack is unlocked to any hardware platform.
Speaker B: What?
Speaker D: A modern embedded stack is unlocked to any hardware platform.
Speaker C: So that one, that one. We need to triple click on that one. Ah. And that's going to be for an Iot coffee talk.
Speaker B: Yeah.
Speaker D: You will love this one. The last one. The only reason you need 5G is because you didn't optimize your system with hai.
Speaker C: Okay.
Speaker B: Okay. Oh, my God. Uh, how long you were. You were. I think you were smoking something when you came up with that one. Yeah, yeah. Uh, you know, yes and no. It's complicated.
Speaker C: I think that's our wrap.
Speaker D: Yeah.
Speaker C: Ending up on these quotes. Everyone, thanks for tuning in, IoT stars, thanks for hosting us and, uh, uh, see you soon, guys. Thanks. Bye.
Speaker B: Bye.
Other episodes covering the same guests and topics, from across The B2B Podcast Index.
Rockwell Automation’s $2 Billion Bet on the Future of Smart Manufacturing - Blake Moret, Chairman and CEO of Rockwell AutomationThe TechEd Podcast · on Digital twins86 / 100
AI in the Network, AI on the Network - with Iain Gillott, WIATelecommunications Industry Therapy · on Digital twins85 / 100
Digital Twins and the Limits of Synthetic Behavior with Olivier Toubia of Columbia Business SchoolData Gurus Podcast · on Digital twins83 / 100
DevOps and observabilityNext in Tech · on Digital twins79 / 100
The Robot Is Waiting on Your Data.AI Proving Ground Podcast · on Digital twins78 / 100
Moody’s x Coupa: Direct spendMoody’s Talks: Risk Reframed · on Digital twins78 / 100
Wi-Fi HaLow by Morse Micro: High bandwidth, long distancer for IoT 2.065 / 100
Software defined connectivity for IoT with Monogoto
Job market scoop: IoT is not immune of fake jobs and candidates
Simplifying IoT Security: Secure Boot, Updates & Edge AI with Window Snyder
The State of Multi-Radio IoT with DP Technics (Walter)