Building the Enterprise Brain with AI, IoT, and Private Data

Podcast John Ward January 21, 2026

So in our inaugural episode of IoT and AI leaders, I've got an old friend of mine, Rob Tiffany. You'll hear Rob describe his background, and it is very varied. As is this podcast, we cover a lot of ground on AI and the positive side of it, the abundance message, and the negative side of it, which is the what could go wrong and what we need to be aware of. Rob has a lot of information from his new role as a research director in IDC, and talks about what he's seeing from large companies and the role. We finish off by talking about the role of the IoT project manager and how it's gonna get more and more important as companies build the enterprise brain as IDC are now talking about it. It's a fantastic episode to kick off the new podcast around the convergence of IoT and AI, and I hope you really enjoy it. Here it is. So before we get started, this is actually a really important episode because for the last four years, we've been doing IoT leaders. And then it became really clear that suddenly the data from IoT is going to be really essential for AI going forward. And some people would say, including potentially my guests this week, that it will actually be more important than the data that's been collected by AI so far. So we're rebranding as IoT and AI leaders. We hope to get another four years worth of podcasts out of this. And my guest this week is Rob Tiffany. And Rob and I have known each other for many years, and we've done a variety of podcasts, some of which we were just talking prior to recording this. Perhaps we wanna forget, but hopefully not this one. We're hoping this one is one we want to remember. And so with that, Rob, welcome to what I now need to call the IoT and AI Leaders Podcast. Excellent. Thanks so much for having me. It's great to be here. Great. And, you know, as I said, we do know each other a while. And because of that, I can actually intro this by saying you have a really interesting background. I mean, everyone has an interesting background, but you have a really interesting background. And it actually starts with a submarine. And maybe we could pick up from there, and maybe you just introduce yourself that way. Yes. Yes. I'm still in the navy driving submarines. No. You're right. That it is kind of unusual. It was a a lifetime ago that in the US Navy, and I was on a couple of submarines. The first one was a seal team delivery vehicle. So, basically, they took some of the old ballistic missile subs from the sixties, took out the ICBMs, where they call them SLBMs on a submarine, to have birthing space for the navy seals. And then on the top flat missile deck, they put the shelter, and they had two mini subs. And so total James Bond kind of stuff. It was really cool. And so did that special ops kind of stuff, and then I was on a Trident submarine, and you have Trident submarines as well in the UK. I do. The the name Trident, comes from the missile, actually. And so in the US, they were called Ohio class submarines. And so not quite as interesting. You're just kinda cruising around really slow waiting for the message from the president. Right. So, yeah, kinda creepy. And and you need to make sure you've not got the Navy SEALs loaded in the tubes. Exactly. A lot of people are like, how you know, I grew up in Texas, they're like, how in the world did you get to the Pacific Northwest and end up at Microsoft and all those places? And literally, I arrived there via submarine and opened a hatch. Here I am because we decommissioned all our nuclear submarines at the shipyard that's across the Puget Sound kind of bay from downtown Seattle. And so there you have it. That's how I got to the northwest just in time for the grunge era to happen in the early nineties. So I really scored on that one. So we we've already covered potentially four podcasts worth of subjects, but we'll narrow it back down to IoT and AI. Because, you know, maybe you could just give a quick thumbnail sketch because I also know you were Ericsson. You've done a you've mentioned Microsoft. You've done a bunch of things. And now you've got a fancy schmancy title at IDC, the analyst. So maybe you could just the listeners and indeed the viewers, we'll see you in that very nice red sweatshirt that you've got on. A quick thumbnail sketch of life since submarines. Life since submarines lately, in addition to being an analyst at IDC covering mostly cloud stuff, I do three times a year ago brief senior officers in the US military, like generals and admirals. I always bring back this IoT thing because it started for me in submarines because submarines, there's no windows. And we use a variety of sensors to tell us our situational awareness. Where are you underwater? What's how fast are you going? Are there contacts and everything? And lots of things keep us alive on that submarine, and it's all based on sensors. And I think that was my first introduction to what would become this IoT thing was this notion of how vital sensor radio radiation sensors, different gas sensors, all kinds of things. And so, yeah, it was there. But I got out of the navy, and I joined a startup in Bellevue, Washington called Real Time Data, and that was in nineteen ninety four, which means I'm older than dirt. And so that's where we kinda dove into and it was it was based on vending machines. And so we took back then, there were no intelligent vending machines. You couldn't swipe your credit card. You couldn't do anything high-tech. There were dumb mechanical put in your quarter, and they kinda had spirals that would push out your candy bars. And sometimes they get stuck. Exactly. And so we had to make dumb vending machines smart, and we used cables inside there. But to do IoT in nineteen ninety four, so it was just re right after the Internet was released to the public. It was ARPANET before that, and it was a DARPA project for the Department of Defense. And so they finally released it for commercial users, and then you saw the explosion of the World Wide Web. Thanks to our good friend down at CERN. And so we started doing that, and we made dumb machines smart. And so I spent time doing there. That's where we got IoT kicked off, and we could dive into that a little more later. After that, I kinda jumped into the whole smartphone revolution because we were obviously heavily involved in early mobile operators back then. Primitive days for cellular. It was pretty bad. And so I was at Microsoft and doing Windows Mobile. And so I was part of that team that launched our smartphone. It derived from the Pocket PC, which was competing against the Palm Pilot. Oh, my mobile and smartphone and then Windows Phone. And so, you know, things were better for us when it was just us competing against BlackBerry. But as soon as the iPhone came along, you had to really get some thick skin Yeah. To deal with that. And so that's probably my first half of Microsoft. Second half was designing and building different parts of the Azure cloud, which was a lot of fun and had lots of stuff around data sync and things like that, and then finished up their designing parts of Azure IoT and deploying that system around the world. And so that was a lot of fun. Got recruited by Hitachi out of that team because I think Hitachi was looking at what GE was doing with Predix and, you know, Hitachi, you know, kinda like a big industrial conglomerate, kinda like Siemens and others. And they wanted that same thing. And all they had was a name called Lumada. And so I thought they had more. But I got recruited out there to come in there and design and build that industrial IoT digital twin platform for them, and that was exciting. And that recall that that was actually based around trains in one of the use cases. Is that right? We did. We did. Absolutely. Absolutely. Big things. Yeah. Big things. Yeah. You know? Because all these Submarines, trains. Submarines, trains, you know, giant windmills creating electricity. And so that was exciting. And as you mentioned, after that, was at Ericsson. I went from spending my time in Japanese factories flying to Tokyo all the time to now I'm flying to Sweden, to Stockholm, to Shista. I was the the VP and head of IoT at Ericsson. And and we had something kinda similar to what you had. You know? It was a was an IoT connection management platform Yeah. That was connected with a bunch of mobile operators around the world. Yeah. So got to go to connect I mean, what a journey. There's so much podcast material. We're gonna struggle to keep this episode short. But right now, let's bring this everybody up to date. So what are you doing now? What am I doing right now? My day to day stuff is I'm as an analyst, I am focused mostly my coverage area that I'm writing about is mostly around the cloud players, public, private, hybrid clouds. You're seeing a lot of people doing multi clouds for a variety of reasons. Sometimes it's because of M and A or something like that, which adds complexity. But it's an interesting time for the hyperscalers Yes. With all this AI thing. Because right now, most of the I'm gonna say ninety nine percent of what's happening in AI right now is happening at these large scale clouds because they're the only ones that have enough compute resources to do it. And so you you know, early on, you saw OpenAI get together with Microsoft. Anthropic has been backed this whole time by Amazon, by AWS. Google arguably should have been the leader the whole time. Obviously, they had DeepMind from the UK. Yes. Yeah. And so but they're I think they may have surged ahead recently with Gemini three. That's looking pretty promising. That's what Sam Altman, everyone you know, people listening to this, and and they will if if they keep up to date, they will have seen Sam Altman. I think he issued a code red. Code red. Code red. And we're falling behind. Yes. I'm not sure what everybody does when a code red comes up that they weren't doing previously. Because my knowledge of these companies is our our daughter works for an an AI company in the UK, but but everybody works so hard anyway and such long hours. And, you know, burnout is a is a real risk for everybody. So what do they then do? Because it does seem like I could imagine sirens going off in the buildings and all time. Elon Musk surge where it says you haven't got eight months. You've got eight weeks. Maybe Right. And Elon Musk is another, obviously, another giant mega player as x AI, which is really interesting, you know, because you think about how long obviously, DeepMind's been doing it longer than anybody in Google, and they were kinda doing research quietly. The real takeaway with OpenAI and ChatGPT coming out three years ago, that originally was supposed to just be out to the public for maybe two or three weeks to just kinda show the public some of the research they were working on. I don't think they had any idea what was gonna happen when people saw what it did. Unleashed a beast. They unleashed a beast, and so I think it caught Google by surprise. And so because they weren't ready to really launch to the public, and then they had to, and they're scrambling. But these people have been researching and working on us forever. You have people coming on OpenAI, like Dario, who goes over to Anthropic. And then you've got here's Elon who has not you know, he was part of OpenAI at the beginning and then left. And he split. Yeah. And he split. He's been focused on rocket ships to Mars and Teslas and stuff like that. What's interesting is how he started from scratch just a couple years ago with nothing and said, I'm gonna build my own competitor to these guys. And it shows how if you have unlimited money in your bank account, you can do a lot. And, again, we will bring it down to IoT. But, you know, Cars and Android robots are IoT. And his recent shareholder session, which I was just watching actually prior to this recording, where he talks about, you know, his opening bid will be a factory in Fremont to build a million Android robots. He had them dancing at the event. People have seen it on YouTube. It's worth watching. But he also talks about the Tesla fully autonomous where within two weeks, which I guess would be pretty soon Yeah. The steering wheel will be taken out of the car, and you will be able to sit and text while while driving. And so Right. I mean, it is all moving so fast. And and that's one of the reasons why we wanted to embed more AI into IoT. And the other one actually is something that that you mentioned and then passed over fairly quickly. Wanted to come back to. Yeah. So I you know, part of my time when I was at Cisco, you know, I was at Cisco fourteen years, and one of my roles was running the cloud program globally, the the cloud strategy. And the reason I mentioned that is that we were always talking about eighty percent of the computing and the data will be stored at the edge. You mentioned edge computing in terms of the cloud architecture. Edge computing. But I've been here almost nine years running an IoT company at SI. But I have to say that it's still very rare for an IoT project to be truly an edge processing. Yeah. We make our own router or router, as you'd say, over there. You know? Be bilingual. Which router we gotta take today? Yeah. And, you know, like in the caustic coffee machines we have it, and that's edge processing. Right? In general, ninety five percent of devices are at least outside of the factory floor are are cellular devices, SIMs and whatever. So we haven't really harnessed the power of edge. At least what I'm trying to say is that I don't really think that cloud edge computing and AI has truly merged yet into some sort of unified architecture. Right. But, you know, in chatting to you and what you're seeing at at IDC, you were telling me that, no, this really is happening. It's not just a prediction about the data coming from the edge and the factory floor and everything. But actually, it is happening, and we're seeing several changes in the cloud, the hyperscalers. But also in these data centers and your concept of the enterprise brain. I mean, there's so much interesting stuff. So let's try and unpack that. Yeah. The edge. Oh my gosh. There's so much to unpack. So much going on at AI speed. So within three, four weeks' time, half of this will be out of date. But let's give it a go. Yeah. I mean, well, just when you talked about cellular Yeah. In in general. You know? Because when I was back at Ericsson, you know, I was trying to be a a counterbalance you know, hundred percent a connectivity company. Right? Ericsson makes cellular kit, and it's all and I remember at the time we were launching five g, and it was a big deal. And the whole world's gonna be on five g and IoT. And remember, we can't gloss over things. IoT has had a rough go of it. It hasn't been the smooth, wonderful path at all. I would say two thirds of all of the sixty episodes that we've done on the on the IoT Leaders podcast, I quote people must be getting fed up with me quoting it. I think I'm gonna make it into a tattoo. It'd be easier. You know, we predict actually, was Cisco, Ericsson, IBM, a bunch of major companies predicted in around twenty ten that fifty billion things will be connected, we went. And then we looked back and said, Oh my God, what did we Never say a number and a timeframe in the same sentence. And we got to about eleven billion, and about eight billion of those were Apple devices. So Right. Yes. It it it was a glorious failure. And I would be this counterbalance to the rest of the leadership team at Ericsson when it came to IoT, because in their world, it's gonna be all cellular. I was like, well, actually, cellular is actually the smallest Yeah. Thirteen thirteen percent of the connectivity. Most IoT actually happens indoors, which sounds strange. I'm not saying it always is always is gonna be like that. But if when we sit in a submarine, in buildings, in skyscrapers, in factories, all those things are indoors. And as you can imagine and this is also a sad thing to think about, and people it's bitten a lot of people. In my experience doing these IoT projects forever, sadly, connectivity was the last thing these teams thought about on their to do list of doing the project. They were more worried about the computers and the data and analytics and all that. Yeah. They thought of connectivity as a commodity. Because they've been hypnotized, Rob, by the cell phone. Yes. They had the device. Bought the cell phone, it had connectivity. Right. And it just worked. And just worked. About the engineering that had gone into the firmware and the board and the and the the back end infrastructure. You and you just had roaming, and roaming worked, and you were never kicked off. You always came home after you've been abroad. And if you were abroad, you swapped the SIM card, and you carried off. The connectivity was like, well, that's the easiest part of the project. Easiest part. And if it was indoors, a lot of people say, well, we're just gonna piggyback on the Wi Fi that's already here. Yeah. And that's what they did. So it's interesting how that played out. And I'd say, well, yeah, cellular is really popular if you're outside and moving around, like connected cars. Ericsson had a lot of connected cars. You know? Shocker. Lots of stuff with Volvo, another Swedish company. Yeah. Yeah. We've been working with AT and T in the US for those kind of things. But, yeah, doors, it was just like, well, we'll use what we got. But then we see the advent of these little moving around robots, one of those AGVs and things like that. Yeah. And going from access point to access point in Wi Fi didn't work as smoothly as we'd like it to. And that's where people started going, I wonder if we should be doing cellular indoors. You know? And so that that came about, and that was the thing. So that's super interesting to see that evolution and people realizing how important connectivity is, you know, because you're right. They haven't thought about it at all. And and would you say actually we had the guy from Volvo, French guy from Volvo on the pod about five episodes ago, and he was Volvo Trucks. And he's responsible for their strategy, and he talked about their goal. We made some sort of YouTube shorts for a ton of views. But one of the quotes he talked about was trying to connect his job description was to connect five hundred million things across a hundred quality manufacturing lines so that the manufacture they could do predictive maintenance so the line never stopped. And that would be worth hundreds of millions to Volvo. So to your point about indoors Yeah. And, of course, yes, he was using private networks, and he was using LoRa, and he was using a whole bunch of things. But, essentially, he had this goal. If you could connect everything, the original vision that connects to twenty ten Yes. Prediction. Whoops. That's right. The Internet of everything. Yeah. The Internet of everything. What we actually got was the Internet of people. But the if you could connect everything, then you could optimize everything, and you could make everything predictive and preemptive. Absolutely. Edge was always gonna be then the apps would move down to the edge, the processing would move down to the edge, and your business processes will be driven like that. And they hadn't turned out, as we know, like that despite the evolution of, you know, clouds, hyperscalers, private clouds. You had an interesting angle when we spoke the last time. I was asking you about these big data centers that being built. You know, they're modest set of players like the one that's being announced, size of Manhattan. There's a small investment, six players racing against each other to control the world. But but the idea of NVIDIA selling their and it's we were talking about, you know, this bubble, another topical subject. When will they pop? When will it pop? Everything pops. Oh, you have a different angle on it because you said, well, you're thinking I'm paraphrasing. You're kind of assuming that these data centers are only gonna be in the hyperscalers. But the idea of the enterprise embracing hybrid AI data centers. Maybe you could open up a little bit on that because that was interesting. Yeah. Because right right now, it seems it almost seems like a consumer play, which is weird. All those big players in AI, they're all in the biggest data centers, and you have a billion people maybe using an app or something to ask questions to to these chatbots. But then there's this other angle, and you and I talked about this shortly after ChadGBT came out. Yeah. A couple of unsuspecting engineers at Samsung accidentally didn't know any better, and they uploaded a bunch of data to ChadGBT about their internal corporate information, private info, and they didn't know it was a problem. And what they didn't know is that everything you upload into ChatGPT will now become absorbed in part of a giant OpenAI LLM. You're giving you're giving your it's The Matrix. You're downloading your brain. That's right. That's exactly So area. Yeah. It was kind of a oops moment there, but a lot of CEOs and a lot of the biggest companies are aware of that. And they and so even though a lot of the big players are saying, no. No. No. Connect all your your things, everything in your enterprise to our LLM, and we promise it's secure and there's a Chinese wall or something like that, and you're gonna be fine. A lot of them aren't necessarily believing that story. And so this notion of private AI running in a company's own data centers on prem in their own colocation facility is a growing thing. You know, because like our numbers at IDC, seventy percent of corporate data is still on prem in a data center or at edge locations or retail outlets, things like that. And so it's not all in the cloud. The cloud players want it to be, but it's not. And so there's this notion of, okay, I've got my private data. I don't want my competitors or other people to see it. I do want the value I'm seeing from these giant LLMs. I want that kind of super intelligence from my company with my specific information. And so the idea there is you can a big a big company and I think it's gonna start with the largest companies. If we talk about really large large, you know, hundreds of billion parameter, you know, open weight models. So I'm gonna have the infrastructure needed, the servers, the GPUs, whether it's NVIDIA or or AMD, you know, and others. I'm gonna have that infrastructure. I'm gonna download a pretrained LLM because there's some it's funny. I was correct, and I a lot of times I'll say an open source LLM. They're not none of them are truly open source because none of them will tell you where the data came from. And some of them might be afraid to tell you where the data came from. Yes. Yes. Well, they don't download. Some people would call it stealing, but that's another Stealing. Yeah. Exactly. They don't wanna get in trouble. So it's like, oh, here's this thing, but you can use it. And so there's different companies and different players that have some open source ones. And so you can download it. It's already pretrained. You run it locally on your infrastructure. And then from there, you want it to be smart about your company. And so that talk around having that enterprise brain, that intelligence about your company. And so anyone who kinda like the poor engineers at Samsung, any one of you who has ever uploaded a PDF to chat GBT or a Word document or an Excel spreadsheet to Claude on Anthropic or whatever, and you notice that it it becomes instantly becomes an expert in all that stuff, and you can ask questions about it. In fact, we do this with IoT when you have a find out a problem with the machine that's gonna fail, and you go, what do I do about it? Well, I uploaded the owner's manual for that machine to ChatGPT, and now it's an expert at it. They can tell me what to do. And so doing that same thing, imagine taking every document, every database table, every everything in the company, and using fine tuning to fine tune that model with that information or rag as well. And all of a sudden, now you've got this intelligence about everything there is to know about your company, and not just now, but in the past. Everything the company ever did, it could be a hundred year old or a two hundred year old company. We talk sometimes about company memory. I especially hear it when old timers and big companies get laid off and they go, wow. You're losing your company memory, all that tribal knowledge about what works and what doesn't. And so now you can imagine, because you always have new employees and new things happening, you can import all that stuff. And unlike just putting it into a normal database, it goes into these things called vector databases. And then all of a sudden, you have this intelligence about your enterprise, and you can ask it any question and find out anything. And so new employees can learn about what happened, and they're making better decisions about business. So let's just pause a second and recap because there are a couple, at least two or three really big subjects that are probably worth a bit more debate that you covered. One is this issue of public private, and are you training the world and your competitors? So the point, I think, just to recap on that, because we're moving pretty fast here. Yes. Back to NVIDIA where we started the conversation NVIDIA and the bubble and our pension Yeah. And everything. What you're saying is don't just think that they're going to be selling to the data centers, but there are going to be corporate equivalents of them next to the manufacturing lines because they don't want that data to be available. The Samsung thing that you talked about, they don't want that data to be potentially available to competitors or nation states, and maybe we'll come back to that as well. So that's interesting because that's exactly the way CloudWinds. I talked about my experience with CloudWinds. So, you know, we're all doomed. AWS, for six dollars a month, you get unlimited compute and storage and and and a Cisco router. That's eighteen thousand dollars so there wasn't great ROI. But then it became hybrid and it always settled down to, as you say, sixty percent, seventy percent still stored behind the BIOS. That's number one. Then you talked about the idea of corporate brain as a competitive advantage. So what if again, I'm paraphrasing back to check my understanding. What if you had solved the security issue? It was behind the firewall, which you've done with general cloud anyway. What if you had the hundred and eighty year history of, you know, every document that had ever Every email, every transcript, every video meeting? I mean, you may not you won't have it back to the same extent the further you go back, but you will have billions of pieces of information behind your firewall. If you could train that, then everyone could be empowered with all the knowledge across the company. That's a very powerful idea that has been completely impossible up until now. And then the third subject you went into was this issue of when people join. I wanna come back to that later on when we talk a little bit about graduate unemployment. But when people join, they would have access to all of that knowledge. Right. Yeah. So that's this is the Gus Hof four scenario we just went through because that says, oh my god. Companies are gonna get more productive. There's an abundance of intelligence. Everything the cost of everything is gonna collapse. Let's say It's like we're creating the library on Trantor from the foundation series from Asimov. The library of everything, easily accessible in firewall, air gapped. Or the mirror or I'll use a I'll use a Harry. I'll trade you that one for a Harry Potter analogy. Okay. What was it called? The mirror of Erised, which is desire backwards. Yes. You could you'd be looked into the mirror, and whatever you wish for would come true. So who would Harry kept seeing his parents. Say that again. Harry kept seeing his parents. He did see his parents. You do have a wonderful habit of taking me off down down down a side street. Sorry about that. He did see his parents. It was very, oh, it was very upsetting, very emotional. Yeah. See. Absolutely. So what so this is the sunny uplands. This is the this is the Gussard fold. This is the amazing potential. Then, of course, on the other side of it, you've got the and I was watching a Steven Bartlett Diary of a CEO episode recently. By the way, his podcast averaged two hours twenty. So you you gotta likely by having by me as a as a as a yeah. But he talked about the six people just in the US who are racing to build AGI. Right. There's no competitive pressure or regulation on any of them to slow down and put the guardrails. You mentioned briefly the security and, you know, Dario spinning out for Anthropic because he was concerned about security, but he's now one of the six racing. Yeah. Then, of course, this even if the six of them agreed, you could argue there's more than six, but broadly, there's six. If the six of them agreed to put a framework on on all of this to protect it, then what about the whole US versus China? Mean, China aren't gonna slow down because the US No. They're not gonna slow down. By the way, two hours before recording this, was watching the news here in the UK. Today is the day you may be aware of this. Today is the day where a country for the very first time banned social media for people under the age of sixteen. Australia. Australia. That's Because they're concerned about there's no guardrails, and there's no mental That's right. And the mental health for children. So Mental health for are trying to do it, but this is a really big subject and probably for another another podcast. So let me take a step back and just look at those three LEGO books, if you like. You've got the fact that we kind of thought and as I said, it's changing so fast. We kind of thought that all of this would be in these huge third party data centers, and it would be like the first instantiation of AWS. You know, the cloud is a computer in somebody else's premise. Right. And it's really low cost, and it's amazing. And and I even if I don't pay pay, I don't pay very much. Then we started talking about the fact that these data centers will go the same way as hybrid cloud did. So seventy percent behind the firewall. And then you talked about the enterprise brain and the future of business processes and companies' competitiveness and knowledge will be enterprise brain, which means it has to be fed by all the data on everything, which include Yes. Things. I I I the IoT things. Absolutely. And the data from things. How much of this is research that IDC are covering, I mean, your role covering? Or you must presumably have a series of clients. I'm not asking you to name them. But Yeah. Everything starts typically with, as you said, with the big companies with who have the most money to experiment. Is this a thing? Do you believe there are a series of companies that are now building the enterprise behind the front world enterprise brain capability? They're they may not have coalesced on all of the concepts behind it. All the big companies are definitely kicking the tires, if you will, that have big infrastructure in the same way that an individual could go to Hugging Face and download a model and try it on their laptop, a small one. As soon as a lot of you know, you always have a lot of a CTO and all the stuff and all these big companies and their tech people. Of course, they were tasked early on. I need you to go become experts at this stuff. And then how can our company use this to empower us? Yeah. And so they've all been kicking the tires. All the vendors out there so if I you know, we know who the hyperscalers are, but then there's the traditional IT players that sell servers, Dell, Lenovo, Super Micro, HP, all those players. Right? They're all working hard to empower these enterprises on prem with liquid cooled servers, including They must see it as a huge opportunity to sell more kit. They do. And in fact, they're like, we're being rescued. Yeah. Because Actually They didn't win the hyperscalers sweepstakes, unfortunately. Sorry. You did win. And and so it gives them a second chance. And we have to give credit to Jensen on this one because Jensen's the one who started talking about the idea of an AI factory initially. And then and you mentioned it. You might have my assembly line here in this real factory, and next door in my data center is my AI factory powering the actual, you know, which also goes back to the IoT stuff, maybe calling it physical AI or that connectivity to the real world. Well, yeah, it is the edge. I mean, it is the data from the edge, which never goes into the public domain. Right. And from a security point of view, massively important given the alternative is perhaps to train the model, which then trains your competitors. Absolutely. And so you no doubt. We survey all these big companies, and we're getting data. But, you know, is it a full blown tidal wave yet? No. But they're all testing it out. They're all trying it out. And then, like all the vendors I described and others, they're all working with customers and they actually come back with case studies and they have actual customers who are doing this. Also, there's value in these small language models too. You know, there's this notion that maybe it's not gonna be the superpower giant know it all LLM. There's a lot of thinking that can be finely tuned with different domains, different expertise, and having small language models that seem to work really well. And I was again, we're all getting our information so rapidly from different sources and podcasts and newsletters and whatever. But but, yeah, what you just said was there was a really interesting one. I don't know what it was in. It was in a, I think, a newsletter I subscribed to. It talks about the difference between the US approach and the Chinese approach to AI. And, of course, it was about a year ago when that new Chinese LLM came out that was, like, thirty times cheaper, and they DeepSeek. DeepSeek, I think. Yes. And it's like, oh my god. And they published all their data, and they've done it with far less compute. And that shocked everybody. And then but recently, what people are saying is that China's approach is actually different. It's along the lines of what you just said. This report was saying that most of the innovation around AI is in vertical niches rather than innovation around AI is in vertical niches rather than general purpose horizontal LLM and agent Right. Than the whole stack Yeah. And the race to AGI. They're actually creating industry vertical versions of it. Now whether it's just the way it's developing or whether or not there's a you wouldn't put it past the Chinese government to have an overall, you know, twenty, thirty year plan on this. They're very Of course. Yeah. That's the way they are. Good at that. Yeah. They did with EVs. But, anyway, yeah, they are actually optimizing AI, not as general purpose, LLMs and general purpose agents and capabilities, but in terms of by business process and by industry vertical. Right. So that's an interesting way because as the as the the race for the LLNs and the people talk about, well, humanity's last exam and all all these benchmarks. But the fact is it's a game of leapfrog, and eventually, everything's just gonna even out. I mean, the difference is are gonna be so minute as to most people say, I don't care. You're gonna ask a question, and a Cisco router is gonna send you right to the proper LLM that has the right data. Yeah. So it'll be abstracted, and you won't see it. But then the idea that how will they differentiate, and it almost like application software, it will become vertical, and it will become horizontal by business process, not just by generic. And talking about application software, again, in your role, is it an IDC? Actually, I I never did ask you your exact typo. I did say the phrase fancy schmancy. What what is It's not that fancy. I'm a research director. Oh, well, that sounds pretty good to me. I'm impressed by it. So that's okay. It's okay. But in your role as research director, at DC, what's gonna happen to the application software business? And let me just lay the table on on so, you know, certainly when I was I I used to be based in Germany and working for HP many, many years ago, and that was when the emergence of SAP. I remember I went to Waldorf. Yeah. Yes. When a it was a small building, it's like a Walmart corporate headquarters. I've been there. Bent Right. Tiny building full of terrified salesman. But I went to Waldorf, and it was just a building with cows in a field behind and Yes. Paso Platner with a vision. And you thought, oh, this German company will never slay the American corporate software companies. And guess what? Guess what? They do. So and now do you think something hey. Mean, if you listen to a company like ServiceNow Which people don't put in the big six, big seven, but actually their focus is very much on agents. And they started off in the ITSM, the IT service management space. And now they're doing these agents all about, you know, automate your HR with agents. They're running adverts where the agent proposes the answer and the human says, yes, modify. Yes. So that raises two questions to try and join the dots here. One is, what is the do you believe is the future of enterprise software? Is there going to be a similar type significant disruption of the enterprise software space by companies creating AI enabled agents. And then after that, we'll get on to the so what happens when the new graduates come out and their jobs come out of university and their jobs get done by agents? But that Right. Let's park that one to the end. What's gonna happen in the enterprise software? The big enterprise software players now I know Oracle racing to start a little bit. Yes. They were building Stargate data center There you go. Near my birthplace of Abilene, Texas out in West Texas near all the oil fields in the Permian Basin because that's where they can get cheap energy. Right? Cheap energy. Yeah. We have Which is all that's always been a that's always been a factor with these data centers. Yeah. Close to hydropower, close to Yeah. Places where the sun shines. That's right. Absolutely. Screwed here, although it does rain a lot certainly at the moment. You know, it's too easy for me to jump on this Gentec AI bandwagon because I remember the summer before last, in the course of one week, all of a sudden, every event I went to, the CEOs of all these big tech companies Said the word agentic. They just started saying agentic. Agentic. Agentic. They didn't even know what they were saying probably. And I've seen the service now. It's good to see Idris Elba the service now. Professor Idris. Yeah. I just love him. And then I see Matthew McConaughey doing the ones for Salesforce dot com for their agent force. And I think what they're doing with the adverts, it's okay to you know, like anything, when you want to invent something, you you have it in your mind, you say it out loud, you maybe you write it down. You haven't really created it yet. And I think they're creating adverts to say, this is what we think. This is the possibility. I don't believe for a second that any of them have fleets of agents doing all this stuff because I've been playing with this stuff too. It's still early days in that space. Now there's that notion, are agents gonna replace apps and application software? That's definitely what I'm getting at. That absolutely. I think it'll be a combination. You know? Because don't don't forget folks. We've had agents for a long time. You could call that a daemon. You could call that an NT service. You could have called that a cron job kicking off a script because agents kind of are invisible software. They're running in the background. They usually are in a loop. And in the past, we've we've had them, and they're deterministic. They're using branching logic if this and that. They have a task to do. And now we're using agents, and they're gonna use that MCP protocol, and they're gonna talk to an LLM nearby. And they have instructions, and they're gonna carry things out for you. And that at least that's the thinking. That's the goal. There's the notion of I'm gonna give them a task. And I have friends who are doing this, and they're doing a bunch of agents to go take on tasks, and they're having success with it. There's also the notion of autonomous agents that just kinda be on from each other, and they'll replicate. Right. Right. We know how that. We've seen that movie before. Yeah. It's And so but that being said, you always have to bring it back to reality. I forgot who it was a friend of mine from Microsoft. He goes, but you know what? Sometimes just clicking a button is a lot more efficient than having an agent do all this back and forth. Each other, and then you especially in I've been playing a lot with AI and not just in preparation for the new spin of the podcast, but just in general because I'm so fascinated by it. And what people don't say very often is that sometimes chatting to either an agent or or just something like ChatGPT. I don't if you remember Fawlty Towers and Manuel, the the the waiter from Barcelona, and he used to Basil Fawlty used to hit him on the head all the time in a train. When he didn't up the stance up, he'd go, okay. K. Honestly, sometimes when you ask Galilein to do something, I just feel like I'm talking to Manuel. I'm talking to Manuel. And and especially it seems especially weak in terms of creating a spreadsheet or a PowerPoint. They don't seem to be able to do that. No. No. But, you know, they are absolutely magnificent at many other things. All of which brings us back to the last question and the big one. Yes. And that is that if you assume with everything going at breakneck speed and all these companies competing, not just the six seven Okay. Companies. I was gonna say Horseman of the Apocalypse, but I'm gonna call them that. Yeah. Yeah. I felt yeah. No. Yeah. That's it. I shouldn't say it for any of our younger audience. They will probably be screaming six seven. We pulled in the younger audience right at that moment when you said six enough to do that. I don't think so, Rob. You're that cool? Probably most people listening to this thing, what on earth are they talking about? I think so. But, anyway, if you assume that the breakneck speed of everything, you know, it's getting better at about forty percent a year. So as I said, you know, imagine in a year's time, we'll think, you know, well, that was quaint what we said a year ago. Right. You know? But one thing that has definitely happened and is is going to definitely have been even more is the work increasingly is going to be done, whether it's by agents or just people using LLMs and requiring less people Is that work is gonna be done. I saw this week Stanford produced a study based on US payroll data, which said that in the last well, this was a data as of April, so it was a little bit out of date already being bigger than this. But it said as of April, back to April twenty four, thirteen, one three percent of the jobs that have are now less jobs have been hired based on US payroll data, which is a pretty accurate source of Sure. And people bandy around the fact that forty percent of graduates are unemployed for at least a year. So I don't wanna particularly ask you about just the generic issue of hiring graduates. But if the model if coming back to your concept of the enterprise brain, which I like. Somebody needs to look at the output from the brain and say, in my experience, as you said earlier, that's a good idea, but it won't work. I I know the model thinks this, but in my experience, just don't go there. Sure. But if you're not hiring the younger people to get that experience Then does mean that the experience gradually retires and disappears from the company? So you actually although you have an enterprise brain, you actually become less competitive rather than more? Where does AI get its knowledge from? It gets it from people and things, right? And if that spigot dries up, that water of knowledge stops flowing, then you're right. It will become out of date. You know, I it's it's early days, you know, I you know, I've talked about it. It's there's the abundance people over here. Yeah. And then there's the doom and gloom people over here. Is it gonna I don't know which sides you know? No doubt about it. You've seen those numbers. I've seen numbers. You know, I spent most of my career at Microsoft in the last two to three years. They've laid off huge numbers of people. Avalanche of layoffs. I know lots of these people. They're the smartest people in the world that I know. And not only are they they they get laid off, they can't find a job to save their life, and it terrifies me. And so that's real. And I know the abundance people are like, oh, you know, don't look at that man behind the curtain of I don't know. Yeah. You know? But that's it really is happening. And, you know, I think, you know, you and I listened to, you know, like, Peter Diamandis or whatever. Peter's Peter's the cheerleader. He's the cheerleader of abundance, but his slightest thing, he goes, you know what? They did a survey, and they're seeing the most of the people in the world are terrified as it turns out. And there is that concern that even if we do get to abundance, there may be an interim period of a number of years that's gonna be a rough road. I think that's the point, Rob. Yeah. And I know we should probably try and bring this to a close because this could become like a Stephen Barber's two hours twenty. We're not sure. But to your point about I think, yeah, my view and this is what we want actually with this new format of the podcast. It's more of a debate. So this is a great opening one. But, yeah, my view is that we're as they say in the London underground, we're in the mind the gap phase because the abundance cheerleaders say a world of of of abundance. Everything is productive. The cost of goods falls. I mean, it's the same as what happened when we outsourced to China. The the cost of goods did fall, but we actually, at the same time, got rid of a lot of the middle manager jobs. Right. So a lot of people ended up being employed, not being able to buy things that were cheaper. A world of abundance and they will always and they say, well, there was always a new job. We know when they go back to the invention of the spinning Jenny, which you call industrial revolution, and then mechanical shovels that got rid of people who dug hole. But there was always there was always more work. There was always That's right. Economy grew. There was always more jobs. Now if you listen to Elon Musk, he says that the physical work, don't think that's immune because of these android billions of android robots, the biggest product in history, as he called it, the other week. And we'll see. I'm skeptical on the I'm skeptical on the They're they are all salesmen. Robot. They are all salesmen, including the robot. I'm skeptical on everyone will have a robot. Right. He even said, well, people will they'll monitor people so they won't we won't need as many prisons because the robots will make sure you're behaving. I mean, I think that's that's definitely sniffing some bad stuff. Yes. But I think the problem will be, yes, ultimately, there will be new types of jobs as there always has been. But as you correctly pinpointed, maybe this is a subject for another podcast. It's the gap. It's the gap. What we're seeing is the effect now of the low level, low hanging fruit being automated, if you like. But we don't see the new jobs yet because we haven't worked out what agents are gonna do. So there's a two, three year period that we're going into, and that's what we're seeing with graduates, for example. Absolutely. And it's not just the graduates. All the talk track in media on the television has been, oh, yeah. It's hurting graduates, and it is. But my experience is I'm also seeing older people who are really smart also losing their jobs as well. And so that's real, and I see it. It's not a fantasy. True. It'll be harder for them to get a job. It will be harder because, you know, there's that ageism thing that always plays in, and they're being paid higher salaries. You know? Well, everyone loves to talk about the different industrial revolutions that we've gone through, and we say, yeah. It's a little painful, but then it creates more jobs, and it's always gotten better. I think the difference is each industrial revolution just took care of one thing, one task that it automated than, you know, like steam, you know, or electricity or whatever. The problem here and so you would upskill, for whatever your job that got replaced. I'm gonna upskill and go up the stack, whatever that means, to do this thing. The problem with AI is it's exponential, and it's so fast that when you try to go upskill to this next job It's gonna overtake you. It's gonna overtake you next week. And the math so that's where I do you have to say Peter Diamandis is a good communicator because the human brain has a problem with visualizing an exponential trend. We can all visualize a linear trend. Yes. It's like here's that famous example where he says, like, if you take sixteen steps and your steps are are are a yard each, then you end up sixteen yards away. If you take sixteen exponential steps, then you circumnavigate the world a certain number of times. But with AI, I mean, it's at least three depending on the benchmark, it's between three and six times better per year. And so there is no way you can outrace it. So I think this calls into I think the bigger discussions will have to be governmental policies are gonna have to come in place because you know how people behave differently when they're put into structures. Social unrest. You have social unrest. Absolutely. We've seen it throughout history. When we start to get to ten percent unemployment, twenty percent unemployment, thirty, pretty soon, you'll have enough people that you think about the pitchforks and torches. Yeah. Yeah. Yeah. Yeah. We're gonna burn society to the ground. I know that sounds crazy and extreme. It it but a lot of people are talking about that. And Yeah. You know, as again, I have to give a note to the to the Bartlett podcast I saw because I'm Yeah. Gaintlessly stealing from it. But he really the guy the the guy interviewed said, we're all getting wound up about immigrants in our countries, but no one's getting wound up around the AI. Just wait. Right. A million a million bots are coming in. Yes. Exactly. Have Nobel Nobel Prize levels. That's right. They're all PhD level. About that. Don't wanna end the don't wanna end the pod on on a on a The trades will help you for a while. Young people, go back to the trades. Go be a welder. Go be an electrician. Be a plumber. Things like that. We still need to build things. It doesn't mean I got my haircut yesterday. Hope for that. If you you can see it Fabulous. YouTube. Thank you. I appreciate your But I was saying to him, you're the last person that I think is gonna be replaced. I'm not gonna get one of Elon's robots to Remember the other saying, the future is here, but it's not evenly distributed. And so you and I and a few others know the bleeding edge that. But if you walk around your city or your town Most most people. Everyone's living their lives, and they haven't been offered it at all. And and, actually, that is a great way to finish, isn't it? Because that's why we are doing what we are doing. We're giving up our time. This pod was always initially, it was about educating people on the power and the opportunity of IoT, and it became very successful. Now, there is a power and the opportunity of IoT to train AI. There's between fifty and one hundred times more data around things than there is around data that's been scraped from the internet. And your point about behind the firewall, private, the enterprise brain, is all going to be happening, and it is the next big thing. And if you're in IoT, that's actually fantastic career prospects and career We can't solve everything or even talk about everything. And, but the areas of the implications and the societal political implications of this are the areas we have to deal with at the same time and the regulation necessary regulation. And I probably just described there another five podcast sets. Podcasts. Yeah. Well done, you. Yes. Yeah. Absolutely. But, you know, I'm I'm hopeful to address this. I'm hopeful for our IoT friends that they've gotten another another parallel lease Yeah. This is their moment. I mean This is their moment. Without exaggeration, I mean, it is it is their the next big wave will be driven by the data of things behind the firewall private, and the people who are gonna get it are the people who work in IoT. And And so you think about doing the normal IoT project. I've got my devices. I'm sending telemetry over some kind of connectivity, cellular from your company. And then I'm gonna land in my edge. I may hop over to a big platform somewhere else, and then I'm doing some kind of analytics or it's going into database. Well, now it's gonna go into a vector database that AI uses, and then it's instantly more easily retrievable by people and bots and everything in a way that we've never been able to do before instead of select star from the SQL database to find out what's going on for the end. Our IoT data is so much more empowered now. And so that's exciting. So you're right. I think the sky's the limit now for IoT these days. It's definitely got a new lease on life. We've got out of that trough of disillusionment Yes. And we're skyrocketing. Well, that's a very positive ending and a great first episode for the new theme. So as always, Rob, you and I could and have talked for hours in in the past. And maybe at some point, I'll invite you back again, and we will revisit this and some of those big issues that we sort of floated towards the end. Sure. Because they are not gonna go away. That's for sure. But listen. Let's end it there. Thanks again for being the guest. Thanks for all the amusing stories. We covered everything from submarines to haircuts, so that was a first. That is a first. You can't good luck finding another pod where you're gonna get that kind of, you know, knowledge and entertainment. Right? So, Rob, thanks for being my guest on The New, and I have to make sure I say it right, IoT and AI Leaders podcast. Guest number one on The New podcast. Thank you. Oh, I'm it's my pleasure. Thanks so much for having me. This has been great. Yeah. Great fun. Thanks.

Overview

AI is moving fast. And most enterprises are not ready for what comes next.

As organizations rush to deploy AI, the real constraint is no longer algorithms or compute. It is whether they have the right data, architecture, and operating model to turn intelligence into outcomes.

IDC Research Director Rob Tiffany joins the podcast to explain why private IoT data is becoming the foundation of enterprise AI:

Why IoT data gives AI real-world context that scraped content never can
The rise of private AI and IDC’s concept of the enterprise brain
Why most enterprise data remains on-prem and what that means for AI infrastructure
How IoT data feeds AI factories, vector databases, and real-time decision systems
Why IoT leaders now sit at the center of AI-driven competitive advantage

Tune in to hear how IoT data unlocks enterprise intelligence and reshapes the future of AI.

Transcript

Intro: You’re tuned in to IoT and AI Leaders, your go-to show for insights, predictions, and big ideas on how IoT is reshaping the world of AI.

Nick Earle: So, in our inaugural episode of IoT and AI Leaders, I’ve got an old friend of mine, Rob Tiffany. You’ll hear Rob describe his background, and it is very varied, as is this podcast. We cover a lot of ground on AI and the positive side of it, the abundance message, and the negative side of it, which is the “what could go wrong and what we need to be aware of”. Rob has a lot of information from his new role as a research director in IDC. And talks about what he’s seeing from large companies, and the role. We finish off by talking about the role of the IoT project manager, and how it’s going to get more and more important as companies build the enterprise brain, as IDC are now talking about it. It’s a fantastic episode to kick off the new podcast around the convergence of IoT and AI, and I hope you really enjoy it. Here it is.

Nick Earle: So, before we get started, this is actually a really important episode, because for the last 4 years, we’ve been doing IoT Leaders, and then it became really clear that suddenly the data from IoT is going to be really essential for AI going forward, and some people would say. including, potentially, my guest this week, that it will actually be more important than the data that’s been collected by AI so far. So, we’re rebranding as IoT and AI Leaders. We hope to get another four years’ worth of podcasts out of this. And my guests this week is Rob Tiffany, and Rob and I have known each other for many years, and we’ve done a variety of podcasts, some of which we were just talking prior to recording this, perhaps we want to forget, but hopefully not this one. We’re hoping this one is one we want to remember. And so, with that, Rob, welcome to what I now need to call the IoT and AI Leaders Podcast.

Rob Tiffany: Excellent. Thanks so much for having me. It’s great to be here.

Nick Earle: Great, and you know, as I said, we do know each other a while, and because of that, I can actually intro this by saying, you have a really interesting background. I mean, everyone has an interesting background, but you have a really interesting background, and it actually… and I don’t know all of it, but all I do know is the bit I know starts with a submarine, and maybe we could pick up from there, and maybe you just introduce yourself that way.

Rob Tiffany: Yes, yes, I’m still in the Navy, driving submarines. No, you’re right that it is kind of unusual. Yeah, it was a lifetime ago in the U.S. Navy, and I was on a couple of submarines. The first one was a SEAL team delivery vehicle. So basically, they took some of the old ballistic missile subs from the 60s, took out the ICBMs, or they call them SLBMs on a submarine, to have birthing space for the Navy SEALs. And then on the top flat missile deck, they put the shelter, and they had two mini subs. And so, total James Bond kind of stuff. It was really cool. And so did that, special ops kind of stuff, and then I was on a Trident submarine, and you have Trident submarines as well in the UK.

Nick Earle: Yeah, we do.

Rob Tiffany: The name Trident comes from the missile, actually. And so, in the U.S, they were called Ohio-class submarines, and so, not quite as interesting. You’re just kind of cruising around really slow, waiting for the message from the president. So, yeah, kind of creepy.

Nick Earle: And you need to make sure you’ve not got the Navy SEALs loaded in the tubes, I guess.

Rob Tiffany: Exactly, exactly. But you know, that’s also, you know, a lot of people are like, you know, I grew up in Texas, and they’re like, how in the world did you get to the Pacific Northwest and end up at Microsoft and all those places? And literally, I arrived there via submarine and opened a hatch you know, here I am, because we decommissioned all our nuclear submarines at the shipyard that’s across the Puget Sound kind of bay in Seattle. And so, there you have it. That’s how I got to the Northwest, just in time for the grunge era to happen in the early 90s. So, I really scored on that one.

Nick Earle: So, we’ve already covered potentially four podcasts worth of subjects, but we’ll narrow it back down to IoT and, and AI. Because, you know, maybe you could just give a quick thumbnail sketch, because I also know you were at Ericsson, you’ve mentioned Microsoft. You’ve done a bunch of things, and now you’ve got a fancy schmancy title at IDC, the analysts, so maybe you could just for the listeners, and indeed the viewers, who see you in that very nice red sweatshirt that you’ve got on give a quick thumbnail sketch of life since submarines.

Rob Tiffany: Life since submarines, life, you know, and I… lately, in addition to being an analyst at IDC, covering mostly cloud stuff, I do, three times a year, go brief senior officers in the U.S. military, like generals and admirals, and I always bring back this IoT thing, because it started for me in submarines, because submarines, there’s no windows. And we use a variety of sensors to tell us our situational awareness. Where are you underwater? How fast are you going? Are there contacts and everything? And lots of things keep us alive on that submarine, and it’s all based on sensors, and I think that was my first introduction to what would become this IoT thing, was this notion of how vital sensors… radio… radiation sensors, different gas sensors, all kinds of things. And so, yeah, it was there. But I got out of the Navy, and I joined a startup in Bellevue, Washington. called Real-Time Data, and that was in 1994, which means I’m older than dirt. And so that’s where we kind of dove into… and it was… it was based on vending machines, and so we took… back then, there were no intelligent vending machines, you couldn’t swipe your credit card, you couldn’t do anything high-tech, they were dumb, mechanical, put in your quarter, and they kind of had spirals that would push out your candy bar.

Nick Earle: I remember, and sometimes they get stuck.

Rob Tiffany: And sometimes they get stuck, exactly.

Nick Earle: Yeah, yeah.

Rob Tiffany: And so, we had to make dumb vending machines smart, and we used cables inside there. But to do IoT in 1994, so it was just right after the internet was released to the public, it was ARPANET before that, and it was a DARPA project for the Department of Defense, and so they finally released it for commercial users, and then we got the explosion of the World Wide Web, thanks to our good friend, and down at CERN. And so, we started doing that, and we made dumb machines smart. And so, I spent time doing there. That’s where we got IoT kicked off, and we can dive into that a little more later.

After that, I kind of jumped into the whole smartphone revolution. Because we were obviously heavily involved in early mobile operators, back then. Primitive days for cellular, it was pretty bad. And so, I was at Microsoft and doing Windows Mobile. And so, I was part of that team that launched our smartphone. You know, it had derived from the Pocket PC, which was competing against the Palm Pilot. Then we had Windows Mobile, and Smartphone, and then Windows Phone. And so, you know, things were better for us when it was just us competing against BlackBerry.

But as soon as the iPhone came along, you had to really get some thick skin to deal with that. And so, that’s probably my first half of Microsoft. Second half was designing and building different parts of the Azure Cloud, which was a lot of fun and had lots of stuff around data sync and things like that and then finished up their designing parts of Azure IoT and deploying that system around the world, and so that was a lot of fun. I got recruited by Hitachi out of that team because I think Hitachi was looking at what GE was doing with Predix, and, you know, Hitachi is kind of like a big industrial conglomerate, kind of like Siemens and others. And they wanted that same thing, and all they had was a name called Lumata, and so I thought they had more. But I got recruited out of there to come in there and design and build an industrial IoT digital twin platform for them, and that was exciting to start.

Nick Earle: And that was, if I recall, that was actually based around trains in one of the use cases, is that right?

Rob Tiffany: We did. We did. Absolutely. Absolutely.

Nick Earle: Yeah.

Rob Tiffany: Big things, yeah, you know, because all these… Submarines, trains, yeah. Submarines, trains, you know, giant windmills creating electricity, and so that was exciting. And then, and as you mentioned, after that I was at Ericsson. And so, I went from spending my time in Japanese factories, flying to Tokyo all the time, to now I’m flying to Sweden to Stockholm, to Shista. And so, I was the VP and Head of IoT at Ericsson, and we had something kind of similar to what you had. You know, it was an IoT connection management platform.

Nick Earle: Yep.

Rob Tiffany: That was, you know, connected with a bunch of mobile operators around the world.

Nick Earle: Now, I mean, what a journey. There’s so much podcast material, we’re going to struggle to keep this episode short. But right now, let’s bring everybody up to date.

Rob Tiffany: Yes.

Nick Earle: So, what are you doing now?

Rob Tiffany: What am I doing right now? My day-to-day stuff is, as an analyst, I’m focused mostly… my coverage area that I’m writing about is mostly around the cloud players, public, private, hybrid clouds. You’re seeing a lot of people doing multi-clouds, for a variety of reasons. Sometimes it’s because of M&A or something like that, and so… which adds complexity, but it’s an interesting time for the hyperscalers.

Nick Earle: Yes.

Rob Tiffany: With all this AI thing, right? You know, because right now, most of the… I’m going to say 99% of what’s happening in AI right now is happening at these large-scale clouds, because they’re the only ones that have enough compute resources to do it. And so, you know, early on, you saw OpenAI get together with Microsoft. Anthropic has been backed this whole time by Amazon by AWS. Google arguably should have been the leader the whole time. Obviously, they had DeepMind from the UK.

Nick Earle: Yes, yep.

Rob Tiffany: And so… but they’re… I think they may have surged ahead recently with Gemini 3. That’s looking pretty promising.

Nick Earle: And that’s what Sam Altman, everyone, people listening to this, and they will… if they keep up to date, they will have seen Sam Altman, I think he issued a code red.

Rob Tiffany: Code Red!

Nick Earle: Code red, saying, we’re falling behind.

Rob Tiffany: Yes!

Nick Earle: I’m not sure what everybody does when a code red comes out that they weren’t doing previously because my knowledge of these companies is… our daughter works for an AI company in the UK but everybody works so hard anyway, and such long hours, and, you know, burnout is a real risk for everybody. So, what do they then do?

Rob Tiffany: Because it does seem like… I can imagine sirens going off in the building.

Nick Earle: It’s like an Elon Musk surge, where he says, you haven’t got 8 months, you’ve got 8 weeks.

Rob Tiffany: That’s right. And Elon Musk is another, obviously, another giant mega player is XAI, which is really interesting, you know, because you think about how long… obviously, DeepMind’s been doing it longer than anybody, Google – they were kind of… doing research quietly. You know, the real takeaway with OpenAI and ChatGPT coming out 3 years ago, that originally was supposed to just be out to the public for maybe 2 or 3 weeks to just kind of show the public some of the research they were working on. I don’t think they had any idea what was going to happen when people saw what it did.

Nick Earle: Unleash the beast.

Rob Tiffany: They unleashed a beast, and so I think it caught Google by surprise. And so, because they weren’t ready to really launch to the public, and then they had to, and they’re scrambling. But these people have been researching and working on this forever. You have people coming out of OpenAI, like Dario, who goes over to Anthropic, and then you’ve got… here’s Elon, who has not, you know, he was part of OpenAI at the beginning and then left.

Nick Earle: Very split, yeah.

Rob Tiffany: And he split, and he’s been focused on rocket ships to Mars and Tesla’s and stuff like that. What’s interesting is how he started from scratch just, what, a couple years ago, with nothing. And said, I’m going to build my own competitor to these guys, and it shows how, if you have unlimited money in your bank account, you can do a lot.

Nick Earle: And again, we will bring it down to IoT, but you know, cars and Android robots are IoT, and his recent shareholder session, which I was just watching, actually, prior to this recording, where he talks about you know, his opening bid will be a factory in Fremont to build a million Android robots. He had them dancing at the event. People haven’t seen it on YouTube, it’s worth watching. But he also talks about the Tesla, fully autonomous, where within two weeks, which I guess will be pretty soon.

Rob Tiffany: Yeah.

Nick Earle: The steering wheel will be taken out of the car. And you will be able to sit and, text while driving. And so, I mean, it is all moving so fast, and that’s one of the reasons why we wanted to embed more AI into IoT. And the other one, actually, is something that you mentioned, and then passed over fairly quickly, I wanted to come back to. So, when I was at Cisco, you know that I was at Cisco 14 years, and one of my roles was running the cloud program, globally, the cloud strategy. And the reason I mention that is that we were always talking about 80% of the computing and the data will be stored at the edge. And you mentioned edge computing in terms of the cloud architecture. Right. Edge computing, and then I’ve been here almost 9 years running an IoT company, Eseye, But I have to say that it’s still very rare for an IoT project to be truly an edge processing. Yeah, we make our own router, or router, as you’d say over there. You know, we need to be bilingual.

Rob Tiffany: Which route are we going to take today?

Nick Earle: Yeah, yeah, and you know, like, in the cost of coffee machines, we have it, and that’s edge processing. But in general, you know, 95% of devices are… or at least outside of the factory floor, or cellular devices, SIMs and whatever. So, we haven’t really harnessed the power of edge. At least, what I’m trying to say is that I don’t really think that cloud edge computing and AI have truly merged yet into some sort of unified architecture. Right. But, you know, in chatting to you and what you’re seeing at IDC, You were telling me that, no, this really is happening, it’s not just a prediction about the data coming from the edge, and the factory floor, and everything, but actually it is happening, and we’re seeing several changes in the cloud, the hyperscalers, but also in these data centers and your concept of the enterprise brain. I mean, there’s so much interesting stuff, so let’s try and unpack that.

Rob Tiffany: Yeah, yeah, the edge.

Nick Earle: Oh my gosh.

Rob Tiffany: There’s so much.

Nick Earle: There’s so much going on, and at AI speed, so within 3-4 weeks’ time, half of this will be out of date, but let’s give it a go.

Rob Tiffany: Yeah, I mean, just when you talked about cellular, in general, you know, because when I was back at Ericsson, you know, I was trying to be a counterbalance 100% a connectivity company, right? Ericsson makes cellular kit, and it’s all… and I remember at the time, we were launching 5G, and it was a big deal, and the whole world’s going to be on 5G and IoT. And remember, we can’t gloss over things. IoT has had a rough go of it. It hasn’t been the smooth, wonderful path at all.

Nick Earle: I would say two-thirds of all of the 60 episodes that we’ve done on the IoT Leaders podcast I quote, people must be getting fed up with me quoting it, I think I’m going to make it into a tattoo, it’d be easier. You know, we predict… actually, it was Cisco, Ericsson, IBM, a bunch of major companies predicted in around 2010 that 50 billion things will be connected, and then we looked back and said, oh my god.

Rob Tiffany: Oops.

Nick Earle: Never say a number and a time frame in the same sentence, and we got to about 11 billion, and about 8 billion of those were Apple devices. So, yes, it was a glorious failure.

Rob Tiffany: And, you know, and I would be this counterbalance to the rest of the leadership team at Ericsson when it came to IoT because in their world, it’s going to be all cellular. I was like, well, actually, cellular is actually the smallest…

Nick Earle: 13% of IoT connectivity.

Rob Tiffany: I go most IoT actually happens indoors, which sounds strange. I’m not saying it always is going to be like that, but when we talk… Right, in a submarine, in buildings, in skyscrapers, in factories, all those things are indoors, and as you can imagine, and this is also a sad thing to think about, and people… it’s bitten a lot of people. In my experience doing these IoT projects forever, sadly, connectivity was the last thing these teams thought about on their to-do list of doing the project. They were more worried about the computers and the data and analytics and all that.

Nick Earle: Yeah, yeah.

Rob Tiffany: They thought of connectivity as a commodity.

Nick Earle: Because they’ve been hypnotized, Rob, by the cell phone, because the cell phone, you bought the cell phone, it had connectivity, and it just worked, and you never thought about the engineering that had gone into the firmware, and the board, and the back-end infrastructure, and you just had roaming, and roaming worked, and you were never kicked off.

Rob Tiffany: Absolutely.

Nick Earle: Right. You always came home after you’d been abroad, and if you were abroad, you swapped the SIM card and you carried on. Yeah, the connectivity was like, well, that’s the easiest part of the project. Right. You always came home after you’d been abroad, and if you were abroad, you swapped the SIM card and you carried on. Yeah, the connectivity was like, well, that’s the easiest part of the project.

Rob Tiffany: That’s the easiest part. And if it was indoors, a lot of people would say, well, we’re just going to piggyback on the Wi-Fi that’s already here.

Nick Earle: Yeah.

Rob Tiffany: And that’s what they did. And you know, and so it’s interesting how that played out. And I’d say, well, yeah, cellular is really popular if you’re outside and moving around. Like, connected cars. Ericsson had a lot of connected cars, you know. Shocker, lots of stuff with Volvo, another Swedish company, you know, and working with AT&T in the US for those kind of things. But yeah, indoors, it was just like, well, we’ll use what we got. But then we see the advent of these little moving around robots, one of those AGVs and things like that.

Nick Earle: Yeah.

Rob Tiffany: And going from access point to access point in Wi-Fi didn’t work as smoothly as we’d like it to. And that’s where people started going, “huh, I wonder if we should be doing cellular indoors?”. You know, and so that… that came about, and that was the thing. So that’s super interesting, to see that evolution, and people realizing how important connectivity is, you know, because you’re right, they haven’t thought about it. At all.

Nick Earle: And would you say… actually, we had the guy from Volvo, French guy from Volvo, on the pod about 5 episodes ago, and he was Volvo Trucks, and he’s responsible for their strategy, and he talked about their goal. We made some, sort of YouTube shorts for it, which got a ton of views. But one of the quotes he talked about was, trying to connect… his job description was to connect 500 million things across 140 manufacturing lines, so they could do predictive maintenance, so the line never stopped. And that would be worth hundreds of millions to Volvo. So, to your point about indoors. And of course, yes, he was using private networks, and he was using LoRa, and he was using a whole bunch of things, but essentially, he had this goal, if you could connect everything. The original vision- going back to the 2010 prediction, whoops.

Rob Tiffany: The Internet of everything.

Nick Earle: Yeah, the internet of everything, yeah. What we actually got was the internet of people, but the, if you could connect everything, then you could optimize everything, and you could make everything predictive and pre-emptive. Absolutely. The edge was always going to be… then the apps would move down to the edge, the processing would move down to the edge, and your business processes would be driven like that. And it hadn’t turned out, as we know, like, despite the evolution of, you know, clouds, hyperscalers, private clouds.

You had an interesting angle when we spoke the last time. I was asking you about these big data centers that are being built. You know, they’re modestly, like the one that’s been announced, the size of Manhattan. A small investment, six players racing against each other to control the world. But the idea of NVIDIA selling their… and is… we were talking about, you know, is the bubble, another topical subject. When will it pop? When will it pop? Everything pops. You have a different angle on it, because you said… well, you’re thinking… I’m paraphrasing. You’re kind of assuming that these data centers are only going to be in the hyperscalers, but the idea of the enterprise. embracing hybrid AI data centers. Maybe you could open up a little bit on that, because that was interesting.

Rob Tiffany: Yeah, because you’re right, right now, it seems… it almost seems like a consumer play, which is weird. All those big players in AI, they’re all in the biggest data centers, and you have a billion people maybe using an app or something to ask questions to these chatbots. But then there’s this other angle, and it all started, and you and I talked about this, shortly after ChatGPT came out.

Nick Earle: Yeah.

Rob Tiffany: A couple of unsuspecting engineers at Samsung accidentally didn’t know any better, and they uploaded a bunch of data to ChatGPT about their internal corporate information, you know, and private info, you know, and they didn’t know it was a problem. And what they didn’t know is that everything you upload into ChatGPT will now become absorbed in part of a giant OpenAI LLM.

Nick Earle: You’re giving… it’s the matrix, you’re downloading your brain.

Rob Tiffany: That’s right. It was kind of an oops moment there, but a lot of CEOs at a lot of the biggest companies are aware of that, and they… and so… even though a lot of the big players are saying, no, no, no, connect all your things, everything in your enterprise to our LLM, and we promise it’s secure, and there’s a Chinese wall, or something like that, and you’re going to be fine. A lot of them aren’t necessarily believing that story.

And so, this notion of private AI running in a company’s own data centers, on-prem, in their own co-location facility, is a growing thing. You know, because, like, our numbers at IDC, 70% of corporate data is still on-prem, in a data center, or at edge locations, or retail outlets, things like that. And so, it’s not all in the cloud. The cloud players want it to be, but it’s not. And so, there’s this notion of, okay, I’ve got my private data, I don’t want my competitors or other people to see it, I do want the value I’m seeing from these giant LLMs. I want that kind of superintelligence from my company with my specific information. And so the idea there is you can, a big, a big company, and I think it’s going to start with the largest companies, if we talk about really large, large, you know, you know, hundreds of billion parameter, you know, open weight models. So, I’m going to have the infrastructure needed, the servers, the GPUs, whether it’s NVIDIA or AMD, you know, and others. I’m going to have that infrastructure, I’m going to download a pre-trained LLM, because there’s some… it’s funny, I was corrected, a lot of times I’ll say an open source LLM. They’re not… none of them are truly open source, because none of them will tell you where the data came from, and some of them might be afraid to tell you where the data came from.

Nick Earle: Yes, yes, well, they don’t download. Some people would call it stealing, but that’s another.

Rob Tiffany: Yeah, exactly, they don’t want to get in trouble. So, it’s like, oh, here’s this thing, but you can use it. And so, there’s different companies and different players that have some open-source ones, and so you can download it, it’s already pre-trained, you run it locally on your infrastructure, and then from there. you want it to be smart about your company, and so that talk around having that enterprise brain, that intelligence about your company.

And so, anyone who, kind of like the poor engineers at Samsung, any one of you who has ever uploaded a PDF to ChatGPT, or a Word document, or an Excel spreadsheet to Claude on Anthropic, or whatever, and you notice that it becomes… instantly becomes an expert at all that stuff, and you can ask questions about it. In fact, we do this with IoT. When you have a… find out a problem with the machine that’s going to fail, and you go, what do I do about it? Well, I uploaded the owner’s manual for that machine to ChatGPT, and now it’s an expert at it. They can tell me what to do. And so, doing that same thing, imagine taking every document every database table, every… everything in the company, and getting it and using fine-tuning to fine-tune that model with that information, or RAG as well.

And all of a sudden, now you’ve got this intelligence about everything there is to know about your company. And not just now, but in the past. Everything the company ever did, it could be a 100-year-old or a 200-year-old company, we talk about sometimes about company memory. You know, especially here when, old-timers at big companies get laid off, and they go, wow, you’re losing your company memory, all that tribal knowledge about what works and what doesn’t. And so now, you can imagine, because you always have new employees and new things happening. You can import all that stuff, and unlike just putting it into a normal database, you know, it goes into these things called vector databases, and then all of a sudden you have this intelligence about your enterprise, and you can ask it any question and find out anything, and so new employees can learn about what happened, and they’re making better decisions about business.

Nick Earle: So, let’s just pause a second and recap because there are a couple, at least two or three really big subjects that are probably worth a bit more debate that you covered. One is this issue of public-private, and are you training the world and your competitors? So, the point, I think, just to recap on that, because we’re moving pretty fast here. Yes, back to, back to NVIDIA, where I started the conversation, NVIDIA, and the bubble, and our pensions, and our… everything. What you’re saying is don’t just think that they’re going to be selling to the data centers. But there are going to be corporate equivalents of them next to the manufacturing lines, because they don’t want that data to be available. The Samsung thing that you talked about, they don’t want that data to be potentially available to competitors or nation-states and back to that as well. So that’s, so that’s interesting, because that’s exactly the way cloud went. I talked about my experience with cloud. It was all, you know, we’re all doomed. AWS, you know, for $6 a month, you get unlimited compute and storage and a Cisco router.

Rob Tiffany: What else?

Nick Earle: $18,000, so there wasn’t great ROI. But then it became a hybrid, and it always settled down to, as you say, 60%, 70% still stored behind the barrel. So that’s number one. Then you talked about the idea of the corporate brain as a competitive advantage, so what if… Again, I’m paraphrasing back to check my understanding. What if you had solved the security issue, i.e. it was behind the firewall, which you’ve done with general cloud anyway. What if you had the 180-year history of, you know, every document, every email, every transcript, every video meeting, I mean… I mean, you may not… you won’t… certainly won’t have it back.

Rob Tiffany: Right.

Nick Earle: to the same extent the further you go back, but you will have billions of pieces of information behind your firewall. If you could train that, then you could have a… everyone could be empowered with all the knowledge across all of the company. That’s a very, very powerful idea that has been completely impossible up until now. And then the third subject you then went into was this issue of when people join, and I want to come back to that later on when we talk a little bit about graduate unemployment, but when, people join. They would have access to all of that knowledge. That’s the glass half full scenario we just went through, because that says, oh my god, companies are going to get more productive, there’s an abundance of intelligence, everything… cost of everything is going to collapse.

Rob Tiffany: It’s like we’re creating the library on Trantor from the Foundation series from Asimov. The library of everything, easily accessible in firewall, air-gapped, it’s all good.

Nick Earle: Or the mirror… or I’ll use a I’ll use a Harry… I’ll trade you that one for a Harry Potter analogy. Okay. What’s it called? The Mirror of Erised, which was desire backwards, which is…

Rob Tiffany: Yes.

Nick Earle: You could… if you looked into the mirror, and whatever you wish for would come true.

Rob Tiffany: Mmm.

Nick Earle: So…

Rob Tiffany: Right? Harry kept seeing his parents.

Nick Earle: Say that again?

Rob Tiffany: Harry kept seeing his parents.

Nick Earle: He did see his parents. You do have a wonderful habit of taking me off down a side street.

Rob Tiffany: Sorry about that.

Nick Earle: Let’s come back. He did see his parents, it was very… oh, it’s very upsetting, very emotional. Anyway, so what… so this is the Sunny Uplands. This is the glass half full; this is the amazing potential. Then, of course, on the other side of it, you’ve got the… and I was watching a Steven Bartlett Diary of a CEO episode recently. By the way, his podcast averaged 2 hours 20, so you got off lightly by having me as a… But he talked about the 6 people just in the US who are racing to build AGI.

Rob Tiffany: Right.

Nick Earle: And there’s no… there’s no competitive pressure or regulation on any of them to slow down and put the guardrails. You mentioned briefly the security and, you know, Dario spinning out for Anthropic because he was concerned about security, but he’s now one of the six racing and then. Even if the six of them agreed. You could argue there’s more than six, but broadly there’s six… if six of them agree to put a framework on all of this to protect it, then, what about the whole US versus China? I mean, China isn’t going to slow down because of the U.S. By the way, two hours before recording this, I was watching the news here in the UK, and today’s the day… you may be aware of this, today’s the day where a country, for the very first time, banned social media for people under the age of 16. Australia.

Rob Tiffany: Australia?

Nick Earle: Yeah, because they’re concerned as there’s no guardrails, and there’s no mental health for children, so…

Some countries are trying to do it, but this is a really big subject, and probably for another podcast. So let me take a step back and just look at those three LEGO bricks, if you like.

You’ve got the fact that we kind of thought, and as I said, it’s changing so fast, we kind of thought that all of this would be in these huge third-party data centers, and it would be like the first instantiation of AWS. You know, the cloud is a computer in somebody else’s premise. And it’s really low cost, and it’s amazing, and even if I don’t pay, I don’t pay very much. Then we started talking about the fact that these data centers will go the same way as hybrid cloud did, so 70% behind the firewall. And then you talked about the enterprise brain. And the future of business processes and companies’ competitiveness and knowledge will be enterprise brain, which means it has to be fed by all the data on everything, which includes things, the IoT things. The data from things.

Rob Tiffany: Absolutely.

Nick Earle: How much of this is research that IDC are covering, and in your role covering, or… you must presumably have a series of clients, I’m not asking you to name them, but everything starts typically with, as you said, with the big companies who have the most money to experiment. Is this a thing that you know, do you believe there are a series of companies that are now building the enterprise behind the firewall enterprise brain capabilities?

Rob Tiffany: Yeah, they may not have coalesced on all of the concepts behind it, but they’re… a lot of… all the big companies are definitely kicking the tires, if you will, that have big infrastructure, and so… in the same way that an individual could go to Hugging Face and download a model and try it on their laptop, a small one. As soon as a lot of, you know, you always have a lot of, you know, CTOs and all this stuff at all these big companies and their tech people, of course they were tasked early on, I need you to go become experts at this stuff. And then how can our company use this to empower us?

Nick Earle: Yeah, but…

Rob Tiffany: So, they’ve all been kicking the tires. All the vendors out there, so if I, you know, we know who the hyperscalers are, but then there’s the traditional IT players that sell servers, right? Dell, Lenovo, Super, Micro, HP, all those players, right?

Rob Tiffany: They’re all working hard to empower these enterprises on-prem with liquid-cooled servers, including…

Nick Earle: They must see it as a huge opportunity to sell more kit.

Rob Tiffany: They do, and in fact, they’re like, we’re being rescued! Because they didn’t win the hyperscale sweepstakes, unfortunately. Sorry, you did win. And so it gives them a second chance, and they’re… and also, we have to give credit to Jensen on this one, because Jensen’s the one who started talking about the idea of an AI factory initially. And then… and you mentioned it, you might have my assembly line here in this real factory, and next door in my data center is my AI factory powering the actual, you know, which also goes back to the IoT stuff, maybe calling it physical AI, or that connectivity to the real world.

Nick Earle: Well, yeah, it is the edge. I mean, it is the data from the edge which never goes into the public domain. Right. And from a security point of view, massively important, given the alternative is perhaps to train the model, which then trains your competitors.

Rob Tiffany: Absolutely, and so you… no doubt, we survey all these big companies, and we’re getting data that, you know, is it a full-blown tidal wave yet? No. but they’re all testing it out, they’re all trying it out, and then, like all the vendors I described and others, they’re all working with customers, and they actually come back with case studies, and they have actual customers who are doing this. Also, there’s value in these small language models, too. You know, there’s this notion that maybe it’s not going to be the super-powered, giant, know-it-all LLM. There’s a lot of thinking that can be, you know, finely tuned, you know, in different domains, different expertise, and having small language models that seem to work really well.

Nick Earle: And I was, again, we’re all getting our information so rapidly from different sources and podcasts and newsletters and, whatever. But, yeah, what you just said was, there was a really interesting one, I don’t know what it was in, it was in a newsletter I subscribed to. It talked about the difference between the US approach and the Chinese approach to AI. And, of course, it was about a year ago, wasn’t it, when that new Chinese LLM came out that was, like.

Rob Tiffany: Hmm.

Nick Earle: 20, 30 times cheaper, and that, you know…

Rob Tiffany: DeepSeek.

Nick Earle: Yes And it’s like, oh my god, and they published all their data, and they’ve done it with far less compute. And that shocked everybody. And then… but recently, what people are saying is that China’s approach is actually different, it’s along the lines of what you just said. This report was saying that most of the innovation around AI is in vertical niches, so rather than having a general-purpose horizontal LLM and agent and the whole stack. They’re actually creating industry vertical versions of it. Now, whether that’s just the way it’s developing, or whether or not there’s a… You wouldn’t put it past the Chinese government to have an overall, you know, 20, 30-year plan on this. They’re very, very good. at that, like they did with EVs. But anyway, yeah, they are actually optimizing AI, not as general purpose LLMs and general-purpose agents and capabilities, but in terms of by business process and by industry vertical. Right. So that’s an interesting way, because as the race for the LLMs, and people talk about, well, humanity’s last exam, and all these… benchmarked. But the fact is, it’s a game of leapfrog, and eventually. What do you think it’s just going to even out? I mean, the differences are going to be so minute as to most people say, I don’t care.

Rob Tiffany: It is.

Nick Earle: What do you think it’s just going to even out? I mean, the differences are going to be so minute as to most people say, I don’t care.

Rob Tiffany: You’re going to ask a question, and an equivalent of a Cisco router is going to send you right to the proper LLM that has the right data.

Nick Earle: Yeah, so it’ll be abstracted, and you won’t see it, but then the idea that… how will they differentiate, and it almost, like application software, it will… it will become vertical, and it will become horizontal by business process, not just by generic. And talking of application software. Yeah. As… again, in your role at IDC. Actually, I never did ask you your exact title. I did say the phrase, fancy schmancy.

Rob Tiffany: I’m a research director.

Nick Earle: Oh, well, that sounds pretty good to me.

Rob Tiffany: That’s okay.

Nick Earle: But in your role as research director at IDC, what’s going to happen to the application software business? And let me just lay the table on that one. So, you know, certainly when I was, you know, used to be based in Germany, and working for HP many, many years ago, and that was when the emergence of SAP, I don’t remember what. I went to Waldorf when it was a small building, it’s like a Walmart corporate headquarters, I’ve been there… A tiny building, full of terrified salesmen. But I went to, Waldorf, and it was just a building with cows in a field behind, and… Paso Platner with a vision, and you thought, no, this German company will never slay the American corporate software companies, and guess what?

Rob Tiffany: Guess what?

Nick Earle: So, and now, do you think something… I mean, if you listen to a company like ServiceNow, which people don’t put in the Big Six, Big 7, but actually their focus is very much on agents. I mean, they started off in the ITSM, the IT service management space, and now they’re doing these agents all about, you know, automate your HR with agents, they’re running adverts. Where people just, the agent proposes the answer, and the human says yes, no, modify.

Rob Tiffany: Yes.

Nick Earle: So that raises two questions to try and join the dots here. One is, what do you believe is the future of enterprise software? Is there going to be a similar type significant disruption of the enterprise software space by companies creating AI-enabled agents? And then after that, we’ll get onto the so what happens when the new graduates come out and their jobs… come out of university and their jobs get done by agents? But let’s park that one to the end. What’s going to happen in the enterprise software… the big enterprise software players? Now, I know Oracle are racing to stay relevant.

Rob Tiffany: They are. Yes, they were building Stargate Data Center.

Nick Earle: There you go.

Rob Tiffany: near my birthplace of Abilene, Texas, out in West Texas, near all the oil fields in the Permian Basin, because that’s where they can get cheap energy, right? Cheap energy.

That’s always been a factor with these data centers.

Nick Earle: Yep. Close to hydropower, close to… Hydro, yeah, yeah, exactly. And, places where the sun shines. That’s right, absolutely. And we’re screwed here, although it does rain a lot, so, certainly at the moment.

Rob Tiffany: You know.

It’s too easy for me to jump on this agentic AI bandwagon, because I remember the summer before last, in the course of one week, all of a sudden, every event I went to, the CEOs of all these big tech companies. They just started saying agentic, agentic, agentic. They didn’t even know what they were saying, probably. And I’ve seen the ServiceNow, it’s good to see Idris Elba doing the ServiceNow.

Nick Earle: Set the word agenda.

Nick Earle: Professor Idris, yeah.

Rob Tiffany: It’s love him, and then I see Matthew McConaughey doing the ones for Salesforce.com for their agent force.

Nick Earle: Yeah?

Rob Tiffany: And I think what they’re doing with the adverts… It’s okay to, you know, like anything, when you want to invent something, you have it in your mind, you say it out loud, maybe you write it down. you haven’t really created it yet, and I think they’re creating adverts to say, this is what we think, this is the possibility. I don’t believe for a second that any of them have fleets of agents doing all this stuff, because I’ve been playing with this stuff, too. It’s still early days in that space. Now, there’s that notion, are agents going to replace apps and application software?

Nick Earle: That’s what I’m getting at.

Rob Tiffany: That, absolutely. I think it’ll be a combination, you know, because don’t forget, folks, we’ve had agents for a long time. You could call that a daemon, you could call that an NT service, you could have called that a cron job kicking off a script, because agents’ kind of are invisible software, they’re running in the background, they usually are in a loop. And in the past, we’ve had them, and they’re deterministic. They’re using branching logic, if this and that, they have a task to do, and now we’re using agents, and they’re going to use that MCP protocol, and they’re going to talk to an LLM nearby, and they have instructions, and they’re going to carry things out for you, and at least that’s the thinking, that’s the goal. There’s the notion of “I’m going to give them a task”, and I have friends who are doing this, and they’re doing a bunch of agents. to go take on tasks, and they’re having success with it. And so, there’s that notion. There’s also the notion of autonomous agents that’ll just kind of be on.

Nick Earle: They’ll learn from each other, and they’ll replicate. Right, right, we know how that… we’ve seen that movie before. Yeah. And so.

Rob Tiffany: But that being said, you always have to bring it back to reality, I forgot who, it was a friend of mine from Microsoft, he goes, but you know what? Sometimes just clicking a button is a lot more efficient than having an agent do all this back and forth.

Nick Earle: I’ve been playing a lot with AI, not just in preparation for the new spin of the podcast, but just in general, because I’m so fascinated by it. What people don’t say very often is that sometimes chatting to either an agent or just something like ChatGPT. If you remember Fawlty Towers, and Manuel, the waiter from Barcelona, and he used to… Basil Fawlty used to hit him on the head all the time in a train, and when he didn’t understand stuff, he’d go, okay, okay.

Honestly, sometimes when you ask LLM to do something, I just feel like I’m talking to Manuel. And especially, it seems especially weak in terms of creating a spreadsheet or a PowerPoint. They don’t seem to be able to do that. But, you know, they are absolutely magnificent at many other things, all of which brings us back to the last question, and the big one. And that is that if you assume, with everything going at breakneck speed, and all these companies competing, not just the 6, 7, big companies. I was going to say Horsemen of the Apocalypse.

Rob Tiffany: 6, 7.

Nick Earle: Yeah, oh yeah, no, yeah, that’s 6. I shouldn’t say for any of our younger audience, they will probably be screaming.

Rob Tiffany: We pulled in the younger audience right at that moment when you…

Nick Earle: I don’t think we’re cool enough to do that. I don’t think so, Rob.

Rob Tiffany: You’re that cool.

Nick Earle: Probably most people listening to this saying, what on earth are they talking about when they say 6, 7? But anyway, if you assume that the breakneck speed of everything, you know, it’s getting better at about 40% a year, so as they said, you know, imagine in a year’s time, we’ll think, well, that was quaint, what we said a year ago.

But one thing that has definitely happened, and is going to definitely happen even more, is the work increasingly is going to be done, whether it’s by agents or… just people using LLMs and requiring less people is that work is going to be done. I

saw this week Stanford produced a study based on U.S. payroll data. which said that, in the last… well, this was a data as of April, so it’s a little bit out of date already, so it’ll be even bigger than this, but it said as of April, back to April 24, 13% of the, jobs, that have, are now… less jobs have been hired, based on U.S. payroll data, which is a pretty accurate source.

And people will band around the fact that 40% of graduates are unemployed for at least a year. So, I don’t want to particularly ask you about just the generic issue of hiring graduates, but if the model… coming back to your concept of the enterprise brain, which I like. Somebody needs to look at the output from the brain and say, in my experience. as you said earlier, that’s a good idea, but it won’t work. I know the model thinks this, but in my experience, just don’t go there. Sure. But if you’re not hiring the younger people to get that experience. then does that mean that the experience gradually retires and disappears from the company? So, you actually, although you have an enterprise brain, you actually become less competitive rather than more?

Rob Tiffany: I mean, where does AI get its knowledge from? It gets it from people and things, right?

And if that dries up, you know, and that water of knowledge stops flowing, then you’re right, it will become out of date. You know, it’s early days, you and I have talked about it, there’s the abundance people over here. And then there’s the doom and gloom people over here. Is it going to… I don’t know which side’s, no doubt about it, you’ve seen those numbers. I’ve seen numbers, you know, I spent most of my career at Microsoft, and last two to three years I’ve seen an avalanche of layoffs. I know lots of these people. They’re the smartest people in the world that I know. And not only did they get laid off, they can’t find a job to save their life. And it terrifies me, and so that’s real. And I know the abundance people are like, oh, don’t look at that man behind the curtain of… you know, but this… it really is happening. And you know, I think, you know, you and I listen to, you know, like, Peter Diamandis or whatever.

Nick Earle: Peter’s the cheerleader.

Rob Tiffany: He’s the cheerleader of abundance, but his latest thing, he goes, you know what? They did a survey, and they’re seeing that most of the people in the world are terrified, as it turns out. And there is that concern that even if we do get to abundance, there may be an interim period of a number of years that’s going to be a rough road.

Nick Earle: I think that’s the point, Rob, yeah, and I know we should probably try and bring this to a close, because this could become, like, a Stephen Bartlett 2 hours 20, we’re not sure, but to your point about… I think, yeah, my view… And this is what we want, actually, with this new format of the podcast, is more of a debate, so this is really great. Great opening one. But yeah, my view is that we’re, as they say on the London Underground, we’re in the Mind the Gap phase, because the abundance cheerleaders say, a world of abundance, everything is productive, the cost of goods falls. I mean, it’s the same as what happened when we outsourced to China. The cost of goods did fall, but we actually, at the same time, got rid of a lot of the middle manager jobs. So, a lot of people ended up being employed, not being able to buy things that were cheaper.

Rob Tiffany: Yes.

Nick Earle: So, a world of abundance, and they will always be… and they say, well, there was always a new job, you know, when they go back to the invention of the spinning jenny, which is. Industrial Revolution, and then mechanical shovels got rid of people who dug holes, but there was always… there was always more work. There was always… the economy grew, there was always more jobs. Now, if you listen to Elon Musk, he says that, well, the physical work, don’t think that’s immune because of these android billions of Android robots, the biggest product in history, as he called it the other week. And we’ll see. I’m sceptical on the robot.

Rob Tiffany: They’re… they are all salesmen.

Nick Earle: They are all salesmen, including the robot. I’m sceptical on everyone will have a robot. He even said, well, people… they’ll monitor people, so we won’t need as many prisons, because the robots will make sure you’re behaving. I mean, I think that’s definitely sniffing some bad stuff. I think the problem will be, yes, ultimately there will be new types of jobs, as there always has been, but as you correctly pinpointed. Maybe this is a subject for another podcast.

Rob Tiffany: Yeah.

Nick Earle: It’s the gap.

What we’re seeing is the effect now of the lower level, low-hanging fruit being automated, if you like. But we don’t see the new jobs yet, because we haven’t worked out what agents are going to do, so… so there’s a 2–3-year period that we’re going into, and that’s what we’re seeing with graduates, for example.

Rob Tiffany: Absolutely, and it’s not just the graduates. All the talk track in media, on the television, has been, “oh yeah, it’s hurting graduates.” And it is, but my experience is I’m also seeing older people, who are really smart, also losing their jobs as well. And so that’s real, and I see it, it’s not a fantasy.

Nick Earle: That’s true. It would be harder for them to get a job.

Rob Tiffany: It will be harder, because, you know, there’s that ageism thing that always plays in, and they’re being paid higher salaries. You know, everyone loves to talk about the different industrial revolutions that we’ve gone through, and we say, yeah, it’s a little painful, but then it creates more jobs, and it’s always gotten better. I think the difference is each industrial revolution just took care of one thing, one task that it automated, and, you know, like. steam, you know, or electricity, or whatever. The problem here… and so, you would upskill for whatever your job that got replaced, I’m going to upskill and go up the stack, whatever that means, to do this thing. The problem with AI is it’s exponential, and it’s so fast, that when you try to go upskill to this next job.

Nick Earle: It’s going to overtake you.

Rob Tiffany: It’s going to overtake you next week.

Nick Earle: And the maths, and that’s where you have to say Peter Diamandis is a good communicator, because the human brain has a problem with, you know, visualizing an exponential trend. We can all visualize a linear trend.

Rob Tiffany: Yes.

Nick Earle: It’s like, he has that famous example where he says, like, if you take 16 steps and your steps are a yard each, then you end up 16 yards away. If you take 16 exponential steps, then you circumnavigate the world a certain number of times. But with AI, I mean, it’s at least 3, depending on the benchmark, it’s between 3 and 6 times better per year. And so, there is no way you can outrace it.

Rob Tiffany: This calls into… I think the bigger discussions will have to be governmental policies are going to have to come in place, because you know how people behave differently when they’re put into stressful situations.

Nick Earle: Well, you get social unrest.

Rob Tiffany: You’ll have social unrest, absolutely, we’ve seen it throughout history. When we start to get to 10% unemployment, 20% unemployment, 30, pretty soon you’ll have enough people that you think about the, you know, the… Pitchforks and torches.

Nick Earle: Yeah, yeah, yeah, yeah.

Rob Tiffany: We’re going to burn society to the ground. I know that sounds crazy and extreme.

Nick Earle: But a lot of people are talking about that, and…

Rob Tiffany: Yeah.

Nick Earle: You know, I have to give a nod to the, to the Bartlett podcast I saw, because I’m shamelessly stealing from it, but he really… the guy he interviewed said, we’re all getting wound up about immigrants in our countries, but no one’s getting wound up around the AI.

Rob Tiffany: Just wait. Right?

Nick Earle: A million, a million bots are coming in that all have Nobel Prize letters.

Rob Tiffany: That’s right, they’re all PhD levels.

Nick Earle: I don’t want to end up… I don’t want to end the pod on…

Rob Tiffany: The trades will help you for a while.

Rob Tiffany: young people, go back to the trades. Go be a welder, go be an electrician, be a plumber, things like that. We still need to build things.

Nick Earle: I got my hair cut yesterday, I hope you can see it.

Rob Tiffany: Fabulous.

Nick Earle: YouTube, thank you, especially for the compliment.

Rob Tiffany: Fabulous.

Nick Earle: But I was saying to him, you’re the last person that is going to be replaced. I’m not going to get one of Elon’s robots to cut my hair.

Rob Tiffany: Also, remember the other saying, you know, the future is here, but it’s not evenly distributed.

Rob Tiffany: So you and I, and a few others, know the bleeding edge of that, but if you walk around your city or your town.

Nick Earle: Oh, most people.

Rob Tiffany: Everyone’s living their lives, and they haven’t been at all.

Nick Earle: And actually, that is a great way to finish, isn’t it? Because that’s why we are doing what we are doing. We’re giving up our time, this pod was always… initially, it was about educating people on the power and the opportunity of IoT, and it became very successful. Now, there is a power and the opportunity of IoT to train AI. There’s between 50 and 100 times more data around things than there is around data that’s been scraped slash stolen from the internet.

And your point about behind the firewall, private, the enterprise brain is all going to be happening, and it is the next big thing. And if you’re in IoT, that’s actually fantastic career, prospects and career employment. We can’t solve everything or even talk about everything, and but the areas of the implications and societal, political implications of this are the areas we have to deal with at the same time, and the regulation… necessary regulation. And I probably just described there another 5 podcast episodes.

Rob Tiffany: Podcasts. Well done, you.

Nick Earle: Yes, absolutely. But, you know, I’m hopeful… We’ve got to address this.

Rob Tiffany: I am hopeful for our IoT friends that they’ve gotten another parallel, at least.

Nick Earle: They have, this is their moment, I mean… without exaggeration, I mean it is. The next big wave will be driven by the data of things behind the firewall, private, and the people who are going to get it are the people who work in IoT.

Rob Tiffany: So you think about doing a normal IoT project, I’ve got my devices, I’m sending telemetry over some kind of connectivity, cellular, from your company, and then I’m going to land in my edge, and I may hop over to a big platform somewhere else, and then I’m doing some kind of analytics, or it’s going into database. Well, now, it’s going to go into a vector database. that AI uses, and then it’s instantly more easily retrievable by people and bots and everything in a way that we’ve never been able to do before. Instead of select star from the SQL database to find out what’s going on for the, you know… Our IoT data is so much more empowered now, and so that’s exciting. And so, you’re right, I think the sky’s the limit now for IoT these days. It’s definitely got a new lease on life. We’ve gotten out of that trough of disillusionment.

Nick Earle: Yes.

Rob Tiffany: And it’s now skyrocketing.

Nick Earle: Well, that’s a very positive ending, and a great first episode for the new theme. So, as always, Rob, you and I could and have talked for hours in the past, and maybe at some point I’ll invite you back again, and we’ll revisit this and some of those big issues that we sort of floated towards the end. Because they are not going to go away. That’s for sure. But listen, let’s end it there. Thanks again for being the guest, thanks for all the amusing stories. We covered everything from submarines to haircuts. So, that was a first.

Rob Tiffany: That is a first. You can’t… good luck finding another pod where you’re going to get that kind of knowledge and entertainment, right?

Nick Earle: So, Rob, thanks for being my guest on the new, and I have to make sure I say it right, IoT and AI Leaders Podcast. Guest number one on the new podcast, thank you.

Rob Tiffany: it’s my pleasure. Thanks so much for having me. This has been great.

Nick Earle: Yeah, great fun. Thanks.

Outro: You’ve been listening to IoT and AI Leaders. We hope today’s insights help you drive smarter, faster business innovation with IoT and AI at the centre. Thanks for listening. Until next time.

Tagged as:
Data to Value Future of IoT & AI Leadership Insights

About the author

Building the Enterprise Brain with AI, IoT, and Private Data

Overview

Transcript

John Ward

Don't miss future episodes

Nav

Login to enjoy full advantages

Please login or subscribe to continue.

Go Premium!

Enjoy the full advantage of the premium access.

Stop following

Cancel subscription