Tobi: Hello, friends. This is the Alphalist Podcast. I am your host, Tobi. The goal of the Alphalist Podcast is to empower CTOs with the info and insight they need to make the best decisions for their company. We do this by hosting top thought leaders and picking their brains for insights into technical leadership and tech trends.
Tobi: If you believe in the power of accumulated knowledge to accelerate growth, make sure to subscribe to this podcast. Plus, if you're an experienced CTO, you will laugh the discussion happening in our Slack space where over 600 CTOs are sharing insights or visit one of our events. Just go to alphalist.com to apply. intro music]
Tobi: Welcome to the Alphalist Podcast. I'm your host Tobi and today I'm super excited that I have my first real industrial IoT CTO or CEO, formerly CTO of Software AG. Tobi: It's Bernd Groß, as I said, former CTO of Software AG, co-founder of Cumulocity, a successful exit and then he bought back his company from Software AG. Tobi: Why? Because AI is changing the physical world now. Turbines, elevators, machines and maybe your architecture too. Tobi: Bernd, welcome to the podcast.
Bernd: Thank you very much. Looking forward to our discussion. Tobi: I typically start like very early in your childhood. And Bernd, you're like the youngest of seven kids. Is that correct? Bernd: That's absolutely correct, yes. Tobi: And and you're like a full stack nerd household as far as I understood. Like, um, your your your brother is also in tech or?
Bernd: Yeah, exactly. So, uh, let's say half half, not full stack, but actually five brothers, one sister and my actually my father is not a tech at all, so he's more financial services business and uh a banker. Let's put it this way. So and actually when you look what what they did uh in banks uh in terms of digitalization, just 20 years ago still, it's amazing how low tech that business is. So I wouldn't call them full stack tech.
Tobi: Okay, okay. Um and you're like the second or the third or like the Bernd: No, I'm actually the last one, the last broad of from seven kids. Tobi: Wow. So you could cheat. Like you observed everyone and then uh like picked picked the strongest choice which was tech.
Bernd: Well, there there is a various jokes about about that, but actually when you are the uh the youngest uh out of seven, you can imagine what's happening around you with I mean, uh at our house, um I remember my my parents, they're very focused on family, on education and we had friends around from my siblings and and there was always busy, there was always something, right? And it was super difficult to find quiet time, honestly. So I went, I don't know that's a kind of a uh funny thing, but actually I went to public libraries to find my actually to find a quiet times uh uh, you know, because at home it was too too busy sometimes, you know, it was hectic.
Bernd: You know, one of the benefits which gave me is I can nowadays, it's still there in a noisy environment, I can concentrate a lot. I have no issue with noisy environments because of my childhood in that large family, yeah. So, but actually to your question when I think you you hinted towards how did I get in contact with technology, with software engineering and so on, yeah. Tobi: Right.
Bernd: Um yeah, exactly. I I took that over from um uh some of my older brothers who looked into that. Actually they built up their PCs and I was just sitting there and and watching and then I became uh interested myself and um first um first PC we had at home was not really a PC but it was called I think what ZX Spectrum. I don't know if you remember that is probably not your not your there was very early days, right? ZX Spectrum, yeah.
Tobi: I I started with a VC20 which is Commodore but pre C64. Tobi: And um I bought it on the flea market. So yes, I'm uh like one generation a few years. No, a few years, a few years behind you.
Bernd: So that that's right. So the Spectrum was like a small device really and you could actually program in basic very very simple and uh you had to connect it to a TV, right? And now uh think about I I mentioned our household with seven kids. There wasn't too much air time for you in front of a TV because any, you know, people wanted to watch TV and not having me sitting there and programming with a static spectrum. So that was quite funny times, yeah and and the monitor was really a TV, yeah. So that was really old, old school and uh old times and uh and and actually Commodoro was also um at the Commodoro in between before going to um self-made PC, yeah.
Tobi: And was that basic programming or what what was that? Like? Bernd: Yeah, yeah, that was basic programming. Nothing really fancy and and so on. I was just playing around and and then soon after, I mean the um uh you know the modem based connectivity came around. So where you started to do some uh um web experience, the first things. I think the first killer application was it then email or I can't remember what was it actually.
Tobi: I think mailboxes was before etc, right? Like I also remember those times when you had like those magazines where you where you had like code listed and you were like basically typing letter by letter. Kind of not really getting the gist of it. Um, one thing that also sparked my attention uh so later, I guess, did you study IT or what what what was your your like how did you continue your note path then?
Bernd: Yes, exactly. I went to um an engineering uh school. So I I studied actually um uh engineering. I'm a what they call diploma engineer, so kind of uh I think that was uh it's like a comparable with a bachelor. Uh if you would call it bachelor nowadays and then um that was my engineering and I started also to work in an engineering uh setup uh in the first years of my career, yeah.
Tobi: I read like an interesting bit that was that uh like Nokia once wanted to hire you and um I mean this is a like podcast where two Germans uh are are like trying to find the right words in English. Um so and that was also one of the reviews I got. Um funny enough. Early times. Um and you were like you weren't hired because of your English or something or?
Bernd: Yeah, yeah, exactly. So no, that that's a that's a true story. So I um I'm actually based I'm a I grew up uh born and grew up in the Düsseldorf region and in Düsseldorf used to be in 20, 25 years ago here in Germany uh the telecommunication hub, you know, before then uh probably now with Munich uh um kind of took over, but before that you had actually like uh Deutsche Telecom headquarter in Bonn, not too far away, then Vodafone, European headquarter in Düsseldorf is still a large uh uh campus uh there. And then because of that, uh you know, Telecom and Vodafone close by the the the cellular so the mobile uh communication industry settled here in the region of Düsseldorf. You had then uh vendors like uh Motorola, but Ericsson, Nokia and many others and they settled in 90 90, 92. So we are talking really early days here. It's a few years ago, right? And because of the um uh digital cellular business just starting in Europe, yeah. The remember the first GSM standard emerged out of North Europe. So the Scandinavian Nokia and Ericsson really um not invented the standard but have been the main driver globally in the GSM standardization body. So long story short, you had a lot of Telco um businesses and now how here in Düsseldorf and I figured um I I kind of felt that was really booming business that made a lot of sense to me and and be actually at that time people were thinking that will be a B2B business, not like today, right? That was really in the old days in the starting days of these cellular communication and I really wanted to then uh join one of these companies, a cool Scandinavian company. I read about them. They had a different way of working, low hierarchy, uh uh cross functional teams, a lot of stuff we are talking today about. Uh actually I I could do do feel uh was part of a Scandinavian culture and they embedded it into the organizations. So I really wanted to join them. And I applied for for an uh interview with Nokia and they uh uh uh um uh invited me for for for an interview and and uh in in that time 92 uh was actually 92, 93, it was really booming. They were looking for hundreds and hundreds of people joining, right? And uh so they they uh uh they invited me to a hotel because they couldn't find meeting rooms. They were all occupied even on the floor people were sitting with a desk on floors. so was totally boom time in that industry and um and so I I actually again uh uh remember I visited the hotel, I drove with the elevator up was six floor hotel in Düsseldorf and it was just a normal room and two ladies from the HR, one a consultant and one HR lady, they were sitting there in the room on shares and I would they asked me to sit on the bed. So I was sitting myself a really small normal hotel room, yeah and I was sitting there and then they interviewed me on the uh and then actually after after half an hour listen, um I think you fit to our company, you we want to hire you. Let's switch now to English and that's your the last thing we really need to check from you. And since I never really was really uh not at school or during um studies really abroad and I had very uh not not very frequent English conversations up to then, I uh I didn't pass that test. I didn't pass my English test so to speak in my first interview. and I really was upset about that because I wanted to join them. So I went back to the elevator and just before I reached ground level, there was a mirror, the door closed, there was a mirror. I looked at myself and I knew I I I made a decision that I go and uh uh uh for half a year or a year to the US uh to to work there and improve my English, right? So, so that was something which I did then finally. Uh two weeks later I was in a plane on my way to San Francisco, yeah. And uh started to work over there um for some time.
Tobi: Cool, cool. And and and then then you had the the idea for for what Cumulocity is today or what Cumulocity was yesterday or when when did that happen?
Bernd: No, no, that happened then much, much later. I mean, we are now talking about 93, 94, right? So exactly. So that was much later. So I what I did is I went to the US. I actually I I um a friend of mine was there. Uh so I had an easy uh you know, arrival. I could sleep there and then and so on. I settled myself and I found a job and then I I I sort of started to improve my English. I actually then applied back to Nokia. So that's the story, that's the twist here. I applied from the US uh while I was there. I applied for for um again uh with Nokia and actually they invited me then and then this uh you know, uh fast forward a year later, right? I had my interview and then finally they hired me. So I started then working really with Nokia uh in 94. So that's the and was my my first uh let's say 10 years in professional career was really uh with Nokia in uh telecommunication business.
Tobi: And was that more like engineering related or what what were you doing then?
Bernd: Yes, in the in the beginning I was really a uh an engineer, right? Uh but I then very fast because the business was so fast growing. I mean, we doubled revenues every year and then we even doubled it every half a year and so on. I mean, this was really boom time in the cellular business, in the mobile uh mobile communication business and so I again then became uh uh two years later product manager, then from that I became a program manager and then so on and so it really um I I really moved very fast out of the engineering more into I would say uh a technical business oriented job, yeah.
Tobi: Few years to the future. Um How how did Cumulocity happen? And um I I think like then also you moved more like towards technical roles like why why did that happen? Uh like I mean obviously like a CTO is in most cases more like a leadership role. But um how did Cumulocity happen and why? Um and why did you sell and um well then next question, why did you buy back?
Bernd: Quite a few questions, yeah. So well with with with Nokia then, um I moved to the headquarter. So at some point of time, um someone asked me, hey, would you be interested to come over as an expert and uh work in Helsinki or actually in Espo, which is next to Helsinki. Um so I did that um uh more than three years I was based in Helsinki and I became um uh quite uh focused on a new innovation. So I was then heading uh a program for introducing new technology, new innovations to our cellular Telco customers globally, right? And so I was actually managing a team out of um Helsinki doing that for some time, but then um my uh my girlfriend today, my my wife, she was actually with me in Helsinki and in in Finland. We decided, okay, after three years uh Northern Europe, really north of Northern Europe, let's put it in this way, is enough. So we went to the UK still with Nokia, uh where I did um where I did um an MBA at London Business School which was financed by by Nokia. So they sponsored me um a lot in my uh earlier years of the my professional career. So I actually did an uh a master of business administration in addition to my technical background and I think that was very good because that helped me not only to um let's say appreciate technology and understand technology, but also to really always ask um why, I mean, why do we have that technology? What is the purpose of the technology? What do we want to achieve? What is the real um real strategy behind that and and so this kind of a combination of an uh engineering role or engineering education and then uh a business education was very helpful for me. Um and what what then happened is uh I have been asked to go to um set up an innovation incubator for Nokia in Mountain View in in the Bay Area. So, so I was actually already like three years in uh Finland, five years in London and I moved actually from there to to the Bay Area to Mountain View and we and that's why the actually the idea of Cumulocity emerged. That's why I'm telling you the the journey here, how it worked really because I was when I had um the responsibility for that incubation uh unit, um we uh incubated uh uh and that was now 2010. So uh 15 years ago, uh we incubated around cloud computing and cloud computing 15 years ago, I mean it always sounds crazy from today's point of view when you think about what happened with the hyper scalers, but 15 years ago was the early days of AWS. Really the early days of AWS and I remember I met um uh Vogel, the the the CTO of Amazon.
Tobi: Werner Vogels, yeah.
Bernd: Exactly, Vogels a few times over there. Um there was really um enthusiastic uh debate about is it going to uh be a big business or not, who's going to outsource the uh data centers in these hyper scale environments. I mean this really early days and and um there was uh in uh yeah, 2010 and um and we decided then um in Nokia, we wanted to turn our telecom software stack and modernize it and uh and and actually um develop it as a cloud ready software. Yeah, so that was very early days in incubation. So we created a a cloud, yeah, I would say a cloud computing competent center so to speak, that would be the wording today, where we created or incorporate a lot of knowledge, we hired external, we had internal people and and we we engineered an architecture um an yeah, the foundation, let's put it this way, the foundation of the next generation of software stack for the Telco uh um uh products in in in Nokia. And uh Nokia um networks. So it's really um the the network division, the the infrastructure division. And and in in addition to the the the core business uh innovation of modernizing the software stack, um we had um um an incubation unit with 10 different um startup like uh uh incubators, right? So we separated them out of the normal business and uh yeah, an early corporate incubator like we do see nowadays still, I think there's still they used to be on vogue later, but I'm not sure if they're still really on vogue anymore, yeah. But but we have been really very early on that. Again, you know, we are talking now uh 2010, 2012 and and we um long story short there I I got I got really into the um uh machine to machine and and IoT uh uh communication staff and then we um uh we spun off uh that business. So I actually um went to the board of Nokia and I pitched it and then we basically got uh an alignment, uh we agreed on the purchase price and then we we did actually very it was a few weeks, it was very fast. We were able to um separate Cumulocity from Nokia, yeah.
Tobi: Ah. Bernd: and and then we uh basically incorporated back here in Germany because our idea uh was with that technology really an IoT platform technology for industrial uh B2B use cases because we we as a team, we have been B2B, so really a B2B software business and and so we felt it makes more sense to uh relocate out from Mountain View uh back in an industrial environment. So we actually moved back to Düsseldorf where you have uh a lot of industrial uh companies uh surrounding, yeah.
Tobi: And uh in a nutshell so you you're the master of management buyouts then, right? Um because you also bought back from Software AG after selling it then again. And um you might have uh more insights on IoT and how it's how it's how it's working um than than me. So and what what was the platform then in a nutshell? Was it like an alternative to MQTT? I guess not. Um so but using MQTT and then device registry etc or what like what do you do with IoT?
Bernd: So the the original idea of the platform was very very straightforward. I would call it IoT device management. Mhm. You know, uh because we saw that we we actually our our business proposition was based on an um on the uh uh on the reflection that we saw uh the hardware prices were falling, right? So to to connect things, the the price of embedded uh uh uh hardware to be able to connect the machine and and wind turbine and compressor and elevator or manufacturing on the shop floor and that the the hardware tool for the modems and the connectivity was was actually going down almost every year 20%, that's what we perceived. And then the the the connectivity cost, yeah, also the cellular connectivity cost was going down uh also heavily, but what didn't happen was uh and in the back office the solution cost and the um the the operational cost to maintain these device or these these uh fleets of machines, that wasn't reducing and we saw that actually increasing and because a lot of system integrators and and software shops did actually then build solutions on top of these connected devices or machines and and that was almost like they reinvented the wheel again and again and again. That's still happening today, by the way. You know, you see a lot of do it yourself development on top of a hyper scalar environment, but you do actually develop a software stack which is already available as a product, for example with companies like Cumulocity or others. So, so in a sense our original idea is still true that we wanted to productize some of these bespoke development and make it available as a platform. And and this uh and and the the first product uh were the first product was really device management like including firmware management, software management over the air. So software firmware management updates over the air was part of that device management.
Tobi: Okay, and and then some sort of like a um like a operating system plugin or service that you had then to run on all the devices that then reports back, does like executes the firmware updates on depending on the platform, I guess, and some monitoring aspect and some communication aspect as well or machine to machine communication?
Bernd: Yes, absolutely. So the whole really the whole life cycle of of that like, but also then because in the IoT architecture what you typically do is it's it's a real time connectivity. It's different to the architecture before. Um let me give you an example, one of our clients here in Germany is Enercon. So wind turbine OEM. So they produce wind turbines and they sell it almost globally. So we have connected about 30,000 wind turbines with them. And um and they're quite actually quite noisy. They have 20,000 data points they produce and so on. So there's a I think uh we we pump uh 50 terabyte of uh of data of data uh a months into the cloud data lake for them and so on. So there's a lot of data collected. But what is what is actually the the the essence of an IoT architecture here is that you have an a real time connected asset you can manage, yeah. So you have the alarms in real time, you have condition monitoring and you have a feedback loop. So you can do also parameter settings and you can do the software and firmware management and and and and and so on. When you compare that to the generation, the technology which was available before, there was really a VPN system where you needed to dial into the wind turbine sort of, you know, like uh uh you know, desktop streaming type of uh applications, yeah. And then you looked at the check the very ugly user interface with these three structure and then you needed to look into is there some problem or not and what is happening and so on. So it was actually no uh there was no uh real time analytics happening in the background at at all. Really just connecting to that wind turbine with a VPN and it was kind of a, yeah, that was the previous generation, yeah. And and today it's very, very um lot of I guess it grew. I guess it grew.
Tobi: And but why why did you then like you you sold uh which I think was then like that also that wish of every founder at a certain point I might uh merge it into something bigger and uh like achieve financial freedom etc. I can imagine. And and then you bought back um after you also bought from Nokia. Um how did that feel and why did you do that? Um was it like emotional, strategic, why or why did you why did you do it?
Bernd: Yeah, so um when we actually grow the business, um um before I I tell you the reason why we sold it, I I actually uh I want to share you with one one one thing what I really learned right at the beginning. So we had a very nice technology, very nice architecture at the beginning uh with a clear API, rest API structure and remember that is 2012, native cloud, multitenancy, highly scalable, uh horizontal scalable, we had really for each tenant we provisioning own database, we use MongoDB, document oriented JSON, uh so we put a lot of new latest technology into it. That's 2012, right? And and and then I but interesting one when we did our first customer meetings and so on, then uh we presented the slides, we did explain what the technology does and so on and then the people, okay, then let's show me. What do you have? So we we didn't actually because we couldn't show anything, we just had APIs, right? So our idea was that people would utilize and call the APIs and utilize our platform, you know, with our domain model, but there was no application, no UI. So we then actually came up with that device management as our first application.
Tobi: So you then built that old school UI with an ugly tree etc.
Bernd: No, then we actually used a very modern uh modern UI, I have to say no ugly tree and and it became a differentiator actually to be perfectly honest because we did really then for these people, it looked like science fiction because it was very modern web UI driven and so on and intuitive uh but um but what what is what was quite an interesting one. Our perception was um almost like we sell it to the developer persona because rest API and a good documentation would be enough. But the reality was that we sold our technology to OT people who wanted to have more than just APIs because they were not software, they were not software engineers, right? They were operational technology people who were like product managers or field service managers or maintenance people and and so on and they wanted more. So we needed then to actually invest into um heavily into our um application and UI framework and and so on. And and uh long story short then yes, uh um then actually we we we did um we did uh in the beginning the first 18 months was super tough, very little uptake. I have to say a lot of pilots, prototyping, MVPs, but no real commercialization and then we won our first customer and then it it actually we kind of we we we felt um you know, after two years, two and a half years, hey, here's something happening, you know, because people are really um subscribing, we had an online uh subscription, free trials and so on and and and we saw also large software companies getting in contact with us and and one of them was Software AG and they asked uh they wanted to uh to acquire us but we we didn't agree uh because I felt it was much too early and it was just we just enjoyed our ride really. And um and then they OEM us. So they took us into their product offering, OEM us and sold us through their sales channel and Software AG um used to be second biggest software uh company in Germany, 5,000 people, 1 billion revenue and really B2B software company, yeah. And um With a big sales team, I guess, right? With a very big sales team, exactly. And um uh so they took us as an OEM and sold it and became quite quite successful. They kind of became a very fast our second biggest partner and then suddenly the third, the biggest partner and so they came back um because they actually now appreciated the software even more because they could see, hey, they can also sell it, they could also be successful with it and the OEM worked. So the the CEO at that time um uh came back, he did uh contact me, he visited me, we discussed and then finally we we agreed um uh to uh to partner with I always say to partner with them, you call it an exit, I call it a a partnership with Software AG which resulted into a takeover so they acquired 100% of the company, yeah.
Tobi: Okay.
Bernd: And and uh and then now I ask you ask um why did that happen? why why did we do that? And um and that happened in 2017 already, eight years ago now, yeah. And and uh at that point of time, we were uh really at the hype cycle kind of still not at the top of the IoT hype cycle or IoT hype cycle, but it's really really growing a lot in that uh dimension and and and people were really looking into industrial internet of things, internet of things being the new big thing, a bit like what AI is today and hopefully we have time to talk about that later on and and uh so my my idea and actually there there were announcements of big players for example Microsoft announced they will invest 5 billion into the IoT market. Google announced their IoT strategy, AWS followed, um uh other platforms at that point of time Gartner analyst told me they have counted 640 or 650 IoT platforms being available globally, yeah. So uh there was really overwhelming what was happening and and since um uh Software G appreciated our software, they they have been they proven proven that they could sell it, they have been successful with it. Um they also knew what it is worse. So kind of fitted, I mean it fit so it was more a strategic decision to say, hey, it's better to team up with Software G now and use our technology. We felt we were a bit at advanced or we because of our early start, we felt we were from a technology point of view ahead, maybe PTC ThingWorks and us could have were kind of uh on that same same level at the time technology wise at least, yeah. And and so we wanted to capitalize on that and we couldn't be available in like 70 countries uh at once like what uh Software G did offer us, right? So they really offered us um uh to continue with our IoT and and and sell it in 70 countries globally at once and and they made an attractive offer to be perfectly frank with you. So it fitted into the uh competitive landscape of uh being squeezed a bit with with the big players and having that opportunity to really uh grow even faster with Software G.
Tobi: And it ultimately ended up not working out or like why why did you then jump off after a few years? Like or buy back after a few years?
Bernd: Yeah, so um no actually it it worked very well, huh. Um um by the way, I then actually left the IoT domain because there was a new CEO coming in at Software AG uh and he asked me to become the CTO of the whole Software AG group at that point of time and so I lost a bit um uh the connection to the IoT business. I looked more towards IT related uh um solutions offering and help to um uh let's say modernize some of the software of Software AG. Um so we we when they acquired us, we have been I think the only commercial cloud product in the entire company. Think about that. So Software G very traditional company and 2017 still, most of the company most of the business was really enterprise architecture with with on prem and private data center deployments, yeah. Uh and and then they used Cumulocity also to to help uh spearheading uh from a business model, from a business transformation, from a software architecture point of view. So that's why I became group CTO and and um and it worked very well. Um why did I buy it back then actually? That only happened uh uh summer last year, so almost uh a year ago and and the reason was really um um first of all, there was an opportunity to buy it back because Software G was acquired by Silver Lake partners um uh here in Germany. So, so Silver Lake partners took over Software AG. I think it was in uh 22 um and took uh Software AG private. So um used to be uh stock traded uh public company and then put that took us private and and uh so um I foresee, I forsaw um an opportunity that they probably will optimize the operation. There there were seven different product units and um yeah, you could you can see it today, they sold five out of the seven. So I kind of anticipated that. I mean it wasn't too difficult to anticipate let's put it in this way. So I anticipated that there's actually an opportunity, yeah, to go back into um into that uh uh IoT business uh with uh selling the business back from from uh Silver Lake partners. So that was really the idea.
Tobi: So typical private equity comes in, wants to optimize business, uh wants to get rid of certain certain business lines that are maybe like too small or whatever, not profitable enough, whatever um or not fitting to the to the core strategy of of of the new Yeah, exactly or even um I mean if you look at Software G today, um they sold uh what methods that used to be the the backend integration business, stream set that used to be kind of a data fabric tool. Um they sold uh they sold actually out of seven businesses already five. There are two left and there are rumors in the market that they also sell these other two. So what they did is really cut it into pieces and sold the pieces.
Tobi: Like Axel Springer, right? Where like all the digital assets were or most of the digital assets were sold um by by KKR. Uh so that that's that's logic and that's lucky lucky you then. Bernd: I would say. Tobi: Like as as long as your plan plays out. Um and what what do you do differently from from since then? Like you mentioned AI um as like kind of a game changer uh for you as well. Um is that like what really motivates you now or like after so many years or what what what
Bernd: So, I mean, what what is quite interesting if you think about when you're working on a on a project for so many years, all these changes what happened technology wise um it's quite a lot with uh Kubernetes with CICD. I mean, there's a lot of stuff you I mean, we we also listen to other um uh other companies, they you always have to modernize technology wise, but but what I also um I I'm a strong uh believer in um in the bridge between operational technology and IT technology, right? You in in simple terms like the physical world. I I do see a lot of benefits that the physical world will be uh virtualized. Uh you know, think about there's a lot of sensor technology happening nowadays. Um uh I mean, on average, um we even we don't know it, but we on average get in contact with almost 30 sensors a day. Yeah. Uh even so um they are simple sensors like you know um like meeting room uh temperature sensors, meeting room occupation or whatever water sensors, light sensor. I mean, there's there's a lot, but it's increasing, continuously increasing, yeah. And and uh so I I do believe there will be the physical world will be automated through digitalization, yeah. And and that is something which is something I I I believe I I always believed in and then um this connected world how it's often referred to is something I I see emerging uh every day it's even getting closer to it and and you see that um with uh with AI as well. I think uh Nvidia um just in January, they they launched uh a foundation model called Cosmos uh which is there open source model to um to uh the physical world. They kind of they kind of see the next gross uh gross engine um really coming from from uh robotics and and digitalizing the physical world. So because Nvidia is looking for data, right? They're looking for for uh more and more data which they can compute with their with their GPUs and and that's what what is happening. There's a lot of attention and focus and that's what I believe as well if you I mentioned the the 50 terabytes of data from a wind turbine maker, right? Uh which we we we generate um uh uh a month. So this is just one example of it and the more of these examples um the more um the more concrete uh use cases are emerging, the more it becomes clear that that AI is just another accelerator for driving the connected world. So it's another motivation because you you can leverage and and automate um uh workflows even more efficiently with with an AI strategy then without an AI strategy and and so that was one of the reasons why we as a management team, not only me, it's a management team, we are five of us, we strongly believe in the opportunity to drive um an IoT uh platform like we used to be into what we call nowadays an AIOT platform. So it's the new term emerging AIOT and we are strong believers in that.
Tobi: That's an interesting term. Um AI is kind of getting commoditized, right? But I think use cases still win. What does that mean from your perspective for for people building platforms like like you do in a way today? How how do you think about that? Like with AI like you you see AI around the corner, it's getting commoditized. How do you how can you still win as like someone who operates a platform?
Bernd: Now, that's a very strategic, very important question and um and I think um what when I uh personally look at it is that the um let's say the the data topic uh has become super important for many of our clients. They understand how important data is. It has been, I mean it's not an old it's an old news, I know, but it it's just becoming more and more important for industrial uh companies as well, right? So uh so in in the B2B context, right? In the manufacturing, um on on the shop floor, uh also OEMs producing machines like I give the example with a wind turbine or with any other type of machines with compressors or even for elevators. One of our clients Schindler connected almost a million elevators and data, how to analyze the data in real time, predict things, but also automate workflows is super important and and what that is what our technology nowadays does. It started with an IoT device management. Yes, and it's still needed, yeah, it's still needed. You need to maintain the life cycle of your uh machines and fleets in the remote places, out out there. Um but you also need to do um need to focus on the data acquisition, um the data normalization, um the the contextualization of the data. So you need to help also structure the data because that is the foundation for your uh training for your AI model training for example in machine learning use cases which we have a lot nowadays, right? And uh so our platform is is is is doing that. So we do a lot of um data acquisition. We keep typically, we we call it a speed layer architecture. We keep the the real time or near real time or a few weeks of data within our platform, but we also do a lot of offloading into a cloud data lake. And and uh um we have actually a real time offloading where you basically you um uh transform data into a parquet or nowadays iceberg um file format with a metadata uh enrichment of a parquet uh file format and and then you store it in a data lake which has become for many of our clients the the let's say the yeah, the foundation for their digitalization strategies and and we feed this data lake with in real time. Um uh which is uh super important because um it's not like three days old, five days old, it's really uh uh I mean, IT real time um uh can be a minute old or something like this, yeah. But uh but it's it's in in near real time uh and and this is our platform proposition nowadays. So it's not just device management still important, but it's also the whole data operationalization, the data ops part and then uh uh another very important aspect, many people forget about that is, if you uh let's say you do machine learning, you train your models on a machine learning. At the end of the day in our business, that means you have let's say 10,000 machines out there. Like an American client of us Solenis, yeah. They have 10,000 machines out there and and then if you train the model, so you need to you need to actually ensure that the latest model is running on your machines. So how do you do that? It's of course again software management. That is our machine again, our our platform again. So we do really the the the loop. So we help collect the data to normalize the data, to enrich the data in a format which is uh really um uh uh uh productive for for data analytics and and machine learning or AI and then we actually help to uh to operationalize these models and insights. So these these IoT architecture gives you these both, yeah, both loops and um a lot of our clients they are coming from a from a one way street. They collected data, they did something, but they couldn't deploy it. So that there was a gap. We call it a mind the gap uh problem. They there was a gap actually to yeah, to to then make use out of your findings and really operationalize your findings and with an with an AIOT platform like Cumulocity, that's absolutely possible to do.
Tobi: I hope that makes I hope I what you're saying. I hope that makes sense. Bernd: Yeah. Tobi: How how much happens on the edge and how much happens in a centralized data center? I guess like most of it happens on a hyper scalar somewhere, right? Um in some centralized instance or some client specific instance or is there anything that happens at the edge on on on your devices on or like at your clients uh near your clients location or data center or where does it happen?
Bernd: Yeah, I think um uh that's a very good point. I think there's a big, big movement or trend towards edge nowadays, yeah. I think uh at least I would call it uh hybrid. So we we used to have um I would say a lot of cloud uh deployments, yeah. But nowadays it's uh it's uh cloud and edge and in some cases it's also really um edge only, yeah. Um and um and in in our um our technology stack we support all of that. We support actually uh small low footprint devices with an open source initiative sinEdge.io. Then we have an our PC, our industrial PC based edge, which is the same software code then which we run in the cloud. So that means that when you for example um uh operationalize or deploy or innovate applications uh on in the cloud and and you have uh you want to uh internationalize and for example expand to to uh regions where um uh cloud operation from US or from Europe isn't easy because of regulatory restrictions etc. then you can easily deploy that at the customer side on an edge uh because it's the same API structure, it's really the same software. Our we call it thick edge is the same as we have in the cloud and then the thin edge. And and edge is is fundamentally increasing, the importance is increasing as also the the amount of data uh is is increasing a lot, yeah.
Tobi: Understood. Um like as you do that for a few years, um how much technical depth is survivable? I guess you have quite you you established or you've built quite a lot of different in different directions and you must have dealt with like a lot of depth on the way uh and a lot of legacy. Uh just through like the different phases you went through, right? Like you mentioned uh like I don't know, microservices, see like Kubernetes, um now AI around the corner, like uh I guess own protocols on IoT etc while there were like others that now kind of um maybe are are the survivors like MQTT, Modbus etc. Like it's like it sounds like a mess where you really must have dealt with a lot and must have also sunset a lot of your products again and a lot of like many many of of your product parts again and pieces again. How do you deal with it? Like how much like is it how much can you can you deal with?
Bernd: Yeah, that's that's the uh what you're just describing is actually the let's say the nightmare of our uh CTO and the in in Cumulocity, they have to he has to deal with a lot of stuff and and you're absolutely right when you um when you think these these OT technology stack is um uh we we support hundreds of different field protocols. It's it's amazing. It's like uh um uh there's very some famous ones with Modbus, Mbus, CANbus, uh but there are a lot of protocols which are not known, but we still have to support them because you need to support them because this is the the language the machine is really speak, the protocol of the machine and we want to get the data of the machine into the cloud so you need today to actually you need to adapt these into an uh in our case uh it's uh often an MQTT uh based uh uh cloud ready um uh connector because these field protocols are not cloud ready, you know, they're not secure enough, etc. You really need to use, I mean MQTT S uh uh as as our main protocol to connect these things and and and but if you think about these field protocols, you need to maintain, you have to uh release cycle that and so on and and so it's not an easy task and uh a few years ago we have um decided to do that through in that domain, specific domain through an ecosystem partnership. So we have what we call device certified partners. Um like about 100 partners globally who have gateways and and and PLCs and and and and all of that um industrial hardware, um IO link type of master units and so on. I mean, there's a lot. And they certify with our uh Cumulocity uh cloud connectivity. So that is basically uh we deliver the software to these devices from a cloud point of view and they are responsible for mapping their proprietary protocol into our uh domain model so that they can actually use uh and and benefit from our ready to use applications at Cumulocity because we need to map it into our uh model, right? And and and so that that is a few years ago so that helped us a lot nowadays. So we have partners as for example a Japanese partner, we have a lot of business in Japan. Um a Japanese partner Takibishi is the name, no one knows it is really an IoT based OT based company, but they support 320 different PLC protocols globally.
Tobi: Sounds sounds painful, but I also guess that you're not able to like offload everything of that that like let's say crazy uh depth also if you like just look at how your platform evolved. Um but maybe next time we have to have a drink before, so you actually tell me the truth.
Bernd: No, no, but you're right. I mean, honestly, yes, we also maintain a lot of for example narrowband IoT adapters, which are part of the Telco standards, 3GPP standard or you have uh OMA, Open Mobile Alliance standards with lightweight M2M because you have a small devices, small edge sensors which you want to connect. It's super important that you have a a protocol which helps you to prolong the battery lifetime and and reduce the the data communication and so there there's a lot of stuff we still um support natively and have to maintain, yeah. Um but then the big the big uh I think the big uh renovation which we did on that software stack, um because when in 2012 we started, I would I mean let's be open here, we started a bit monolithic because the architecture was at that point of time, a cloud-based uh architecture which was actually from today's point of view would look very monolithic, yeah, one big uh piece of software. And with uh we then introduced um Kubernetes in 2016, 17, before we got acquired, we started actually to deploy. We have been one of the first commercial um uh commercial deployments in Kubernetes, yeah. Um and what we did with it, we we started to cut our software into uh microservices pieces. So we have been really um putting a lot of effort into uh into re-engineering and and and and so on and this is to some extent still happening, yeah. But out of the monolithic software we started, we now have about 30 plus microservices uh in the backend uh running on on our on our Kubernetes cluster. Um so we really modernized and and cut things into pieces and and I think that was a um we didn't know would it be the right strategy or not, but in uh it was the absolutely right thing to do because after that we did then uh continuous integration, continuous deployment. If you wouldn't be able to do that, if you wouldn't be able to split your organization software engineering team into these microservices teams, um so that they can actually deploy the software. Uh so yeah, but it's still I I think we have still a lot of technology depth, that's for sure. That's for sure.
Tobi: Yeah, and also microservices can lead to a lot of trouble as well. But yeah, like we all know that as of today. Uh we we're all wise and and old. Um I I still have a little surprise for you like as an as an outro question um because we like talk quite a lot and super interesting stuff you your you you have on the in the backpack. But um let's imagine um the former Nokia CTO gave me a device like back in the 90s um that he invented, they never make it to made it to like market readiness. But imagine um they wanted to back in the days invent the time machine. Um and they were successful. They never were able to release that because like beforehand the the mobile um mobile phone market crashed really uh for them uh because the iPhone etc. And um imagine like we can now dial a year and we dial 1994 and that was the year when you were in LA meeting the Nokia folks and you just returned from that meeting like super super frustrated because of of the the English uh problem, let's say. Um and you look in the mirror right now, we observe yourself, you look in the mirror um and um yeah, you just like ask yourself like, hey, why why did that happen? Um and you now would have the chance to whisper something into your young younger self ears, uh younger self's ears. What what would that be?
Bernd: That's a that's a it's a very complicated time uh time travel you just indicated but uh but um uh that's uh it's uh it's a good question. So what should I have known at that point of time that which I know now. It's uh there's a lot. There's there's super there's there's a lot. There's a lot. I mean and um uh honestly, I think there's I think the the most for to me personally, the most important um lesson is is not really the the technology stack is how to deal and and manage and and work with people. That is my most important lesson over the years. Um I um because I I grew up with my professional career in this Scandinavian background um which was low low hierarchy. I remember in Espo, I was based in Espo, I mentioned it earlier in the fourth floor, the the CEO of Nokia was just a few tables away. He was sitting in an open desk environment. That is really uh now um 25 years ago, right? The Yoma Olla was his name, the CEO of Nokia and he was sitting just five desk away in an open open uh environment and and that's that's um um my my my kind of uh learning from that was um to be uh uh to be actually to um to gain uh authority through knowledge and competence and not through hierarchy and titles. That's a very simple simple thing, but um it's easy to say, but that's something I I learned many years and and in order to get that, to gain these, um I think you have to to be a trusted person and and and that's the foundation for everything and um and that means you should be just yourself. And I I mean that's easy to say, you know, I know that some business environments where you uh is maybe not as easy as in the IT business, right? I I I appreciate that. Um but that really um in the first years of my career, I think I was more, I was too too too too focused on on how I I I I perceive, what I'm doing. I wasn't always myself, right? I wasn't always, I was kind of feeling that would weaken me. But actually it's the opposite. It's it strengthens you as a leader if you can be uh very open, very direct and and and and and and uh and do not hide anything because in the mid and long run this is becoming visible and people start not to trust you anymore and and I think that has been a a learning it took me 15 years, 20 years of my career to really appreciate and learn that uh and um yeah, that is uh that is uh important to me is really open frank communication and setting up a trustful environment.
Tobi: Yeah, um I I think that took took took many traditional CTOs and CEOs a while to really like internalize that, right? Like this like that leadership is changing and that you like rarely lead through authority.
Bernd: Exactly, exactly, yeah. I mean, you uh you mentioned to me before as if I may ask you a question. You mentioned to me uh that you are also uh investing into uh SaaS startups etc. And I think uh it's probably for you the same. What what you do is you look into the leadership teams, right? As a key indicator. Uh uh it's the same. It is in a in a way, um when you uh when you do that uh and you you learn to read people, you will identify easily if this people is open and and and authentic or is this people this person actually uh playing a role which is actually in the mid and long run not paying off.
Tobi: Yeah, yeah. You you you start seeing features in people like Bernd: Exactly. Tobi: Hey, this Bernd guy. Bernd: Oh, yeah, you got me, you got me. Exactly, yeah. Tobi: Yeah, thanks thanks a lot. Thanks a lot. Great discussion. Um was really really nice talking to you finally and um I wish you a nice day in in Düsseldorf then. Bernd: Yeah, same uh same to you in in Hamburg or in Berlin, where are you at the moment? Tobi: I'm I'm in Hamburg.
Bernd: I wish you a nice day in Hamburg then and um so if some of the um audience want to um get in contact with us at Cumulocity or with me personally, I'm easily reachable through LinkedIn, yeah, just as a small advertising here at the end, yeah. Tobi: Yeah, then you see that he's at Cumulocity and investing into I don't know, 20 different VC funds or something. Bernd: Not so many, not so many, but a few, a few. Exactly. Tobi: Big big investor. Thanks a lot, Bernd and have a great day.
Bernd: Have a great too. Uh thank you very much. Bye.
Tobi: Thank you for listening to the Alphalist Podcast. If you like this episode, share it with friends, I'm sure they love it too. Make sure to subscribe so you can hear deep insights into technical leadership and technology trends as they become available. Also, please tell us if there is a topic you would like to hear more about or a technical leader whose brain you would like us to pick. Alphalist is all about helping CTOs getting access to the insights they need to make the best decisions for their company. Please send us suggestions to [email protected]. Send me a message on LinkedIn or Twitter. After all, the more knowledge we bring to CTOs, the more growth we see in tech. Or as we say on Alphalist, accumulated knowledge to accelerate growth. See you in the next episode.
outro music]