Podcast #223 Data Driven Enterprise

 
TRANSCRIPT
 
Ken: Good day, and welcome to episode 223 of our Momenta Digital Thread podcast series. Today, I'm greatly pleased to host Dominik Obermaier, CTO and co-founder of HiveMQ, a German company that is the most trusted MQTT platform. Dominik co-authors the book "The Technical Foundations of IoT." He is a member of the OASIS Technical Committee and is part of the standardization committee for MQTT 3.1.1 and MQTT 5.0. He is also a frequent speaker on IoT, MQTT, and messaging. Dominik holds a Bachelor of Science in Informatics from the Hochschule Landshut in southern Bavaria. Dominik, welcome to our Digital Thread podcast.

[00:01:25]
Dominik: Hi, Ken. I'm excited to be with you today.

[00:01:28]
Ken: I'm excited to have you too. I've watched you guys' background. We're investors in the space, and I think we had an early opportunity to invest. Sadly, we passed. But I'll put you under 'the one that got away' bucket, if you will. But we were always impressed with the work that you've done, especially the standardization work, to get back in that regard. We call this the Digital Thread podcast, and we always like to start by asking about your digital thread. In other words, the one or more thematic threads that define your digital industry journey. What would you consider to be your digital thread?

[00:02:02]
Dominik: That's a great question, Ken. I think there are multiple aspects of that. If you look back at the arc of my career and the things that started before my career that caught my interest, I was always deeply fascinated by the disruption we make with technology. I'm a big believer in technology; it's one of the ways to push humanity forward. I don't only mean digital technology. I also mean like- you must build technology. Since we exist, this is what we do. Humans build technology. Especially these days, and the last- since the internet emerged, since distribution got like a marginal cost of zero, we see more and more disruption happening faster than ever. There are also other technologies I'm sure we'll also discuss, like AI technologies and others, that will increase the trend even further. 

I'm deeply interested in that. What's happening is disruptions happen. Usually, it's happening gradually and then, suddenly. This is something I have always been fascinated by. There's a second thing: I realized that open standards is one of the ways, especially in computing, to push the envelope even further and especially, faster. One of the marvels of today's world is that we can have this conversation now; our listeners can listen anywhere they are based on technology and open standards. We sometimes forget that like all the boring plumbing of the internet, of all technologies we fill- these are happening on openness. I also believe a big disruption will happen in most industries because open standards will win eventually, even in industries where you don't see a lot of openness. This is something I was always fascinated by, and this was also one of the reasons why I found the company. Since then, I have thought a lot about disruption. I think a lot about improvements over time, but all hopefully lead to an exponential curve. Yes, this is a thread I follow just watching my own career.

[00:04:03]
Ken: I'm always impressed. I liked the digital disruption and the open standards footing you have right there. It does a pretty good job of describing your digital thread. Now, I always like to go back to the origin story, and for you, I looked at your university site. It particularly said that in 2012, you were honored with the best bachelor's thesis in southern Bavaria. What was the focus of that thesis?

[00:04:28]
Dominik: The focus of the thesis was quality assurance in web applications. I have always been very deeply interested in making software that is resilient and of high quality overall. This is also something I did when I started my career. In the first few weeks of my career, I was part of a rollout in an automotive factory in Germany, and I saw firsthand what is happening if production lines go down. Companies are losing a lot of money. I couldn't believe back then, at the beginning of my career, how much money is lost per minute in these factories if they have a downtime. I was deeply impressed, on the one hand, by how stressful it is to recover from the downtime. 

However, I was also surprised by how little time some companies put into high-quality software. Even today, software development pushes a lot for high velocity, which is great, but at the cost of resiliency and high quality. There are certainly some areas where you don't need high-quality software, but overall, especially if you're interacting with the physical world, high-quality software is important because you're letting your users down and, worse, because you're letting your users down with a crappy user experience. This could be one thing for some software, but if you're interacting with the physical, like in a factory, people are actually losing money, companies are losing money, and goods cannot be delivered on time, so it's a big problem. My thesis was also about that, and it also was very interesting. I didn't think about it intentionally back then. But also, if I look at my arc, quality has always been very dear to my heart, and even today, with what we do here at HiveMQ, we focus heavily and invest heavily in quality. We cannot disappoint our users and customers because of what we do in our company. Our software runs 24/7, 365 days a year, so there's no way that we can let our users down anytime. We don't make any compromises on that. My work back then in university was all around this and more focused on web applications. One of the things that I heavily focused on was automation. Even today, I still believe that quality assurance is not automated enough. Humans make mistakes, which is what we do because humans are good at doing creative work. But humans aren't usually as good as machines when it comes to algorithmic work. Especially regarding quality assurance, many of the problems you have here are algorithmic and not creative problems, which I focused heavily on early in my career.

[00:07:16]
Ken: Looking at your background, what intrigued me was your early software development work. You've already mentioned your manufacturing work. You worked across a manufacturing system integrator and a data center provider- both you would consider high reliability and resiliency use cases in there. You already mentioned some of your key insights from that time. Let's fast forward to HiveMQ since you already talked about that. You guys found it in 2012. What inspired you and your co-founders to start the company?

[00:07:48]
Dominik: Yeah, there are multiple things. Number one was when I studied, and also, before I studied, I always worked. I worked in different software companies and, before that, on construction sites; I did a lot of physical work when I was a bit younger. My co-founders and I found out that we didn't find the work environment we were looking for in other companies; this was number one. The second thing is, as a founder, you must be a misfit; you must have some ideas you're unwilling to compromise on to make it happen. This is what we did. We found a company, and since we started, we didn't have a product; we didn't have customers. You have nothing when you start a company. But this isn't something that we learned along the way. Also, our company is different from many traditional startups in the sense that for all of us, this is our first, and all of us were engineers, so there was no real business background. Which was interesting, and the company has a very engineering-driven culture, of which I'm very proud. We have a very customer-focused and almost customer-obsessive culture when solving very hard problems for our customers. Coming back now to why we founded this company, why we founded HiveMQ. Back then, we found early articles from Gartner and Forrester and other analysts who were mentioning and making predictions that in 2025, there will be 50 billion devices connected. 

Later, IDC said, "Oh no, there will be 75 billion devices connected to the internet," and so on. Everybody was talking about a massive quantity of connected things. But how do you actually integrate all of this? What kind of technology would you use to connect billions of devices? Because as a reminder, we now have around 8 billion people living on Earth. In 2012, a bit more than 7 billion people were living on Earth. If you look at the data centers, their computers, and everything we have built so far to support a fraction of humanity connected to the internet, that's just crazy. The web technologies we have built with HTTP and other communication technologies are probably unsuitable. Our hypothesis was that you need a different set of technologies for the Internet of Things and for the Internet of Humans. Our work in the data center showed us that NET application with hundreds of 1000s of users requires so many resources. Things go up in flames all the time. We are not thinking about millions of users, let alone billions, so we didn't see how you would do that. We were under the impression that this needs a paradigm shift. People told us it was a very bad idea; there's no use needing that. Why would you need that? But now, fast forward more than 10 years later, we're very happy that our hypothesis was right. A paradigm shift happened around 2016, bringing us into a good position because we had the product ready to disrupt the market. This is how we currently are. I'm very happy to go into the details of the technology shift we have seen and what Hive brings to the table there.

[00:11:07]
Ken: That's interesting; you didn't have anything when you started: first-time founders, engineers, some ideas, etc. But what impressed me is your background in software development and cross-manufacturing in data center providers. You had invaluable experience that you've already brought in insights, and that certainly speaks to why HiveMQ has done so well. You've really stuck to that as your journey. Mind you, you guys co-founded this in 2012. That was a very different world back then, as you started talking about all the predictions and claims there. You've described the past 10 years at HiveMQ as 'Helping customers build the data foundation and insights they need to create new connected products, find new efficiencies for automation, and scale their businesses within the demands of a real-time communication environment with MQTT.' That is a mouthful, but over 10 years, it sounds like you guys have been quite busy. How are you doing that?

[00:12:05]
Dominik: Yeah, so what HiveMQ was doing is we were building the central nervous system for our world. If you think of the human central nervous system and what it provides us, it's a mixture of decentralization and centralization. We are connecting everything from devices up to things that we now call the Edge, like factories, from machines in the factories to applications in factories, to factories themselves, to other factories around the globe, onto the Cloud. What we provide is this kind of backbone of data transmission that allows you to connect things in real-time and with real-time, I mean, in this case, the fastest time possible, so you can literally, in milliseconds, exchange data at a very, very big scale across the globe. What we do here is we provide a different way of doing things. It's not an incremental change; it's a change to this- what our customers describe as a paradigm shift. We completely changed the way they approach specific problems because now they finally have the ability to connect machines, applications, and other humans on any scale. Imagine it like a message bus if the central nervous system analogy doesn't work for some listeners. Central message bus from Edge to Cloud with multiple hubs so you can, at any given point in time, filter data, aggregate data, and exchange data, and, more importantly, integrate new systems as you go without any integration effort. We found this kind of approach very natural, but also remember in 2012, AWS was only four years old. 

EC2, which is the elastic compute unit that AWS provides, as an example, was available, but a lot of the services on top are not available yet. It was also very early on. This was before Docker was even a thing, way before Kubernetes. But also, back then, we already had the hypothesis- okay, there will be elasticity in workloads. If we think about the Internet of Things deployments, you have actual users using things, so you need this kind of elasticity of workloads. This is where Cloud computing came in. But also, you always need to integrate Edge. Back then, people had different terms for what we now call Edge. But what we now call Edge computing is something we have already focused on. It's like a Cloud-to-Edge conversion, which we focus on very early on. It turns out that going into the manufacturing space is also very helpful because there's something called OT/IT conversions, Operations Technology, and Information Technology conversions. Their companies are almost forced to work together between the digital world and the physical world. This is also where our technology suits a lot these days. Before 2020 or 2018, most of our customers were in the automotive space, connected devices space, so connecting things over the internet. These days, we also do a lot of manufacturing, logistics, transportation, renewables, oil and gas, and other energy options.

[00:15:07]
Ken: What's interesting is when you mentioned your customers, I keep returning to your dual footing in the high-reliability data centers and manufacturing starting up. Because as I look, some of the customers I see are BMW and Daimler on one side, the OT companies, and Netflix and Sirius XM on the other, enterprise IoT. What would you consider some of your notable wins with these clients and perhaps beyond?

[00:15:31]
Dominik: There are a lot of things we've seen over the years, and what's always exciting me is what our customers do for their customers. What I'm always so excited if I look at the customers we currently work with, what they're about to build for their customers, and sometimes we see three years before the customer literally, before their customers see because we're usually brought in very early on in the development of new products of new projects, and so on within our customers. These days, we do a lot of work, especially with Fortune 500 customers and big enterprises around the globe. We focus mainly in the US and also in Europe. This is where we have seen the strongest traction so far. The industry spends from- as I said, from transportation logistics to automotive, to manufacturing, to also like these digital products and services. There are different challenges; for example, if I work with a big automotive company, they have the problem of massive scale because they produce millions of devices each year relentlessly, so you need technology that supports that growth. This is one thing, but if I look at a manufacturing use case, usually, there is a pretty static workflow. It's not like 1000 machines show up one day in a factory, and then you're surprised by additional integrations there. You have much more stale workflows, but other things are getting more important, like reliability, predictable latencies, etc. 

Our technology focuses on catering to all of these extremes, making it very versatile. This is why we work with very different industries. But those are reference customers. As I said, you mentioned some of our reference customers there, but if you look at Fortune 500 companies, there are many of them, especially if you look at the Fortune 100. We work with many of them. What is interesting is that many consider- since we're offering them a paradigm shift of how to do things, many of them consider the work we do as very strategic and a competitive advantage. To give you a few examples, if you get a package delivered somewhere on the globe, chances are HiveMQ is being used. If you buy a European or American car, you will use HiveMQ while you use the device, or at least when it was produced. I think the footprint of what our company does is you see a lot, but usually, you'll never hear the name HiveMQ because we're not customer-focused. We're enabling our customers for that, so this is why we don't have a very strong brand when it comes to consumers, but in the B2B world, especially in the communications world, I think we are pretty well-known at this point in time.

[00:18:05]
Ken: Yeah, I would certainly say so. I'm curious: how do you know when a potential client is ready to adopt your solution? What have you seen as some of the best practices in those clients realizing the full value?

[00:18:18]
Dominik: When a client's way is very different, and there are different models, also what different companies use, but there is a very traditional model, where markets are also segmented in innovators, the early adopters, and then also the early majority, late majority, and laggards. This is a very traditional model. Depending on the industry, we see there are different appetites. In the automotive industry, for example, the technologies we provide are not innovative anymore. This is early maturity, what we see here. It could even be on the verge of late maturity there. 10 years ago, clearly, we worked with innovators; BMW is one of them, but also others, especially German car manufacturers. But these days, MQTT as a technology and HiveMQ as an implementation of the technology are just table stakes for many of these customers, making it easier for us to have conversations with other vendors who are also trying to catch up. 

You see a lot of catch-ups there. When you look at the manufacturing space as an example, we completely see the flip side. There are new ways of accessing data being proposed, especially listeners who work in that space. They might be familiar with the term 'unified namespace' as an example, which provides a concept of how to make data accessible across the whole organization, OT and IT data accessible, and have a central data hub. This is also very HiveMQ; it's being used a lot. But here, we clearly see innovators and early adopters doing this, so I think we're very far from the depiction of the market. Companies are looking into that because there's a lot of pressure from the market and competition to change. It's so interesting. 

A while ago, I read an article that argued that if you look at the S&P 500 over history, I think that more than 90% or even 95% of all of these companies that were once listed on the stock market don't exist anymore. They go out of business because they're only- as cheesy as it sounds, the only constant is change. While a company could get away from being resistant to change for a very long time, until they get disrupted, disruption happens at a much faster pace because distribution costs are zero with the internet, and compute costs are pretty much zero these days. Compared to the past, it's getting better. Energy prices are getting cheaper if you look at any metric there. There's so much that’s going to happen; all of the ingredients for disruption are happening right now. The basic knowledge is there. I think big corporations, especially, are pressured to change, and they're looking for ways to do that. HiveMQ is one of the technology vendors- we usually come in when companies are ready for new projects to try out, get a small dip in change. We have some Fortune 500 companies, especially in the US, who go all in and with us, get business results in a year. They're surprised with the results because a big change doesn't always mean it's very hard to do once you're committed and once the organization really wants to do that. This one, I said- especially big companies, but also a lot of SMBs we work with. We usually don't work with many startups these days, but we also have free offerings. For startups, all our software is free.

[00:21:48]
Ken: Speaking of disruptive technologies, certainly high on most people's minds would be gen AI, aka ChatGPT, at least as people would see it. To what degree have you seen gen AI impacting industrial systems, particularly HiveMQ?

[00:22:05]
Dominik: Yes, interesting question. Many conversations I have with our clients these days are around this topic. Gen AI, per se, might be a bit exaggerated. In the short term, it's overhyped, but in the long term, it's underrated. I think this is the situation you're in. Gen AI is- if you look at the definitions of it, it's just a subset of ML or machine learning with some very specific models and problems to solve. It's also a new field, like one of the foundations for that, the Transformers papers, came out in 2017. It's pretty new, but every company asks themselves, "What does it mean for us?" The thing is, I don't know. If I could predict that this would be really great, I don't know. But there are a few things I do know. Number one, I do know that we are now on this AI journey. I think, as humanity, we haven't even started seeing what it means. I don't even think we have the technology spilled right now that actually will be the winning technology for anything around AI in particular. 

On the other hand, what I do know is if I work with LLMs- which I don't think for manufacturers will be the silver bullet for many of the problems they're looking to solve. I know that if I go to LLMs, traditional machine learning approaches, or other things that haven't even been invented yet- they all need data. They need data access in real-time from anywhere around the globe. This is what we do at HiveMQ. We deliver the data from anywhere around the globe, to anywhere around the globe, at a scale from- let's say, one packet per second up to millions of packets per second, or even more. I think Hive will be the provider for a lot of the data. These days, data is saved on addresses in data lakes and other things. I believe the future will be data in motion that is used to make real-time decisions. Because if you have AI systems that can make decisions based on data that they observe at this specific moment, we don't need to wait for humans. What I know is we will be the main provider, the central nervous system we provide for our customers. We will make them make better business decisions. I'm not convinced that generative AI is the solution for all of the problems there, but it will definitely play a big role.

[00:24:26]
Ken: Great insights. I always like asking the question because we invest in many industrial infrastructure companies. You mentioned UNS or Unified Namespace. The Kepware founders founded a company we've invested in, and most recently, we've invested in Aperio as well, so it's definitely an interesting space. Speaking of the future, you closed an impressive $43 million Series A round in mid-2022. Very impressive, especially for a European headquartered company. What can we expect to see from HiveMQ over this next year?

[00:24:58]
Dominik: There's a lot to see. We scaled the business now nicely with that. We, as a company, are a customer-driven business. We recently released some new products this year. For example, HiveMQ Edge takes the central nervous system even further, down to machines and many proprietary protocols for customers to build a decentralized, unified namespace. This is one of the things we released, and we do it in an open-source way, so we do not believe that integrating proprietary technology should cost money for customers. I think customers should pay money once to get value out of something, and I do not believe plumbing in itself provides value. We are going to market there, so we have some products there. We heavily invested in our own Cloud. 

Many enterprises these days use HiveMQ Cloud as a managed service for connecting their millions of truck drivers for millions of package deliveries, people, and so on. This is what we do, and we also have a lot in store for next year. There are some things I cannot mention right now at this point, but everybody interested in that can follow us very closely on this one. But also what we release is something called Data Hub, which makes it possible for big-scale industrial deployments, but also automotive deployments or any other industry deployments to get insights into data, stop bad data, heal bad data, and finally solve the problem of the garbage in, garbage out problem. That is a bit prohibitive, also for AI technologies, to be very frank. We did a lot of work there. But we haven't even started yet. What we'll release to the market- I think the customers haven't even seen what will be possible going forward next year. I'm very, very excited about this. We believe in open centers. As a company, it's not our job to lock in customers with proprietary technology. Our job is to provide value for customers. Many of the decisions we make are not guided by lock-in principles but by how you can provide more value while customers have the freedom of choice. I'm very excited about next year, so I hope people stay tuned.

[00:27:14]
Ken: Awesome. When you're not co-founding a great startup, writing books on IoT, and serving in a leadership capacity in OASIS standard, where do you find your personal inspiration?

[00:27:26]
Dominik: Far away from technology. I believe technology is a way to drive things forward, but there are some things that never change. Technology is one of the things that always change. This is the vehicle we've built or invented as humanity. We invent technology to drive things forward, but there are some principles that apply, a part of human nature that has been there forever since humans existed. This is something I am very interested in. As an example, philosophy is something. But there's also nature. I believe people today would be much better off spending more time with nature or unplugging from technology. As humanity, we need to get better at using tools correctly, including technology, so I spend a lot of time away from technology when not working. But realistically, I spend too much time with technology based on work. I read a lot of books, listen to many podcasts, and get some inspiration there, but I also ensure a lot of deep conversations with leaders in the industry and in our spaces.

[00:28:31]
Ken: Excellent. One may argue that your time away from technology makes you a better technologist, and your track record speaks to that. Dominik, thank you for sharing this time and these great insights with us today.

[00:28:43]
Dominik: Thank you, Ken.

[00:28:44]
Ken: This has been Dominik Obermaier, CTO and co-founder at HiveMQ, the most trusted MQTT platform- although I would say you guys should use the tagline, 'The Resiliency Company' because that's what you do. Thank you for listening, and please join us for the next episode of our Digital Thread podcast series. We wish you a momentous day. You've been listening to the Momenta Digital Thread podcast series. We hope you've enjoyed the discussion, and as always, we welcome your comments and suggestions. Please check our website at momenta.one for archived versions of podcasts, as well as resources to help with your digital industry journey. Thank you for listening.


[The End]