10 years of IoT trends: What has really changed?

Hi Knud, hi Tobias, great to have you here today. Knud, how do you feel being in the podcast studio? How’s it going for you?

Knud

Hi everyone, and welcome to Hamburg – gray and rainy as always.

We’re practically in your neighborhood, Knud.

Knud

Absolutely! I’m looking forward to our conversation. Let’s get started.

Exactly! For those who are hearing Knud for the first time today, we’ll introduce him shortly. Your company, IoT Analytics, is based here in Hamburg, right?

Knud

Yes, exactly.

We’re currently recording on the Reeperbahn – a place some of you might be familiar with.
Hi to you too, Tobias. How are you doing?

Tobias

Hello Madeleine. I am also doing very well. The short distance from Hanover to Hamburg is ideal.

Awesome. You two already knew each other before, right? But would you mind briefly introducing yourselves? Knud, why don’t you go first? Who are you?

Knud

Hi, I’m Knud. I founded IoT Analytics ten years ago. I’m an industrial engineer and originally came from consulting before starting my own company. For the past ten years, I’ve been fully dedicated to market research – initially on my own, and now with a team. As the name suggests, we’ve been focused on IoT from the beginning. Over the past three to four years, we’ve expanded into adjacent areas, of course, based on customer demand.

A little about myself: I’m originally from Hamburg – it’s my hometown, and you can probably hear it.

Definitely! Why did you found IoT Analytics? We’ve known each other for a while, but I’ve never actually asked you that.

Knud

On a personal level, I wanted to do something I enjoy. I’m an analytical person, I like to dive deep, and I didn’t want to switch consulting projects every three months. I wanted to focus on a single topic. Once I got started, I really enjoyed it and received great feedback. One thing led to another, and suddenly, ten years had passed.

Impressive! You’ve grown significantly over the years. I’m really glad to have you here today – it’s not easy to get you on a podcast.

Knud

Well, that can be interpreted in different ways.

Alright, let’s transition to Tobias. Tobias, tell us a bit about yourself. Who are you, and what’s your background?

Tobias

I’m Tobias Mühlnikel, also from northern Germany – but this time from Hanover.

Shout-out to your hometown!

Tobias

I lead the Edge business unit at Portainer, a New Zealand-based company. Our main focus is on IoT and Industrial IoT, specifically on application management – how applications can be deployed and updated. I also have experience in industry, including at Volkswagen, where I managed smart manufacturing projects.

What exactly did you do there?

Tobias

I worked at the Wolfsburg site, co-leading digitalization projects, including everything related to digital production platforms – a joint venture with AWS and Siemens – and implementing various projects and use cases.

Nice! We might come back to that later because the practical perspective is especially relevant for this episode.
Knud, you mentioned that you’ve been in the field for ten years. Let’s talk about the market. IoT has surely evolved over that time. Looking at the period from 2015 to 2025, what developments or changes have you seen since founding IoT Analytics?

Knud

Back in 2013, there was a McKinsey study on the next big things, and IoT was highlighted as one of the major future trends by 2025. Initially, we were in a hype phase. In the early years, everyone was trying to understand how this new connected world would work. Today, it seems obvious—whether it’s the microphone in this studio or many other things—but back then, it wasn’t. We thought: Wow, the world is becoming interconnected!

In the early years, companies primarily focused on developing architectures—most of it was built in-house because there were hardly any ready-made tools available.

Then, around 2015, the IoT platform phase began. Suddenly, everyone wanted to become the “Android of IoT”—the central platform where all devices come together. Companies from various industries jumped on this trend, claiming to offer THE platform. As a market research firm, we observed this closely and counted the numerous players.

Yes, I remember that time well—at Siemens with MindSphere, it was the same. The number of platforms was overwhelming.

Knud

Exactly! Back then, it was completely unclear who would come out on top. Today, we see that the market has segmented, but in the end, the major hyperscalers took the largest share—something that was far from obvious at the time.

This platform phase lasted from around 2015 to 2018. Eventually, it became clear that it wasn’t just about platforms, but rather about concrete use cases and end applications.
And that’s where IoT Use Case came into play—it fit perfectly with the market’s development.
With the onset of the COVID-19 pandemic, a consolidation phase began. The expectations were huge—50 billion connected devices were predicted, but this growth didn’t materialize as expected. The market continued to grow steadily but not as explosively as many had anticipated.
In the past three years, the market has consolidated in many areas, although there have been new and successful developments along the way.
In the last two years, the topic of AI has changed significantly. While it was always present, various surveys now show how companies are prioritizing it. In 2020 and 2021, AI ranked at the bottom of priorities in such surveys. Today, however, it’s almost at the top or even the number one priority. We are currently in a phase where AI is coming more into focus. The focus is on how IoT can serve as a data source for AI and how IoT-related use cases can benefit from AI. The interplay between the two technologies is becoming increasingly important.

Tobias

So that means the market is slowly shifting from IoT analytics to AI analytics.

Knud

Yes, that’s something I’ve heard more than once.

This is also a major topic for us, particularly the question of where the intersection lies: When do you start working with live IoT data and running AI algorithms on it? And when do you rely on existing databases?

Abstracting these questions isn’t easy—especially in terms of relevance for formats like this podcast. AI doesn’t always run in real-time on IoT data; much of the data is already stored in data pools.

Nevertheless, your latest study shows that the IoT market reached a volume of $270 billion in 2023, even though growth has slowed down.

Knud

Yes, the growth has slowed down as the technology market has matured. However, 2024 was a particularly weak year, especially because the manufacturing sector—where IoT has a strong presence—declined in Europe. This had a direct impact on IoT budgets.

When breaking down the market by technology, we can see that the hardware segment has shrunk—significantly fewer devices were sold. On the other hand, the software segment continued to grow, as reflected in the numbers reported by many IT companies, including Portainer.

Absolutely! It’s fascinating to see where exactly the revenue is being generated. IoT is a multi-billion-dollar market—but is it driven more by hardware or software?

Knud

That’s not so easy to answer because IoT isn’t clearly defined and is therefore difficult to delineate. We look at IoT from the entire technology stack, from sensors to the application layer—excluding legacy applications.

Hardware remains the core, covering sensors, gateways, and various computing devices. However, cloud, AI, and cybersecurity within the software sector have been growing significantly for years.
Another crucial aspect is services. IoT isn’t something that can always be implemented with a plug-and-play approach overnight.

Yes, but luckily, not everything is standardized—many things still need to be developed individually in projects.

Tobias

Despite the dominance of the major hyperscalers, who hold significant market shares, the market remains highly fragmented. An interesting report from your company last year showed that there are still around 3,300 active IoT startups in 2024. This highlights that a significant market still exists, and consolidation is happening gradually.

Knud

There are numerous lucrative niches because IoT touches almost every industry and offers countless use cases. Many companies prefer working with smaller teams of 10 to 30 people rather than large corporations, which keeps the market highly fragmented.

Tobias

How do you see the difference between IoT and Industrial IoT in market observation? These areas are growing at different rates, aren’t they?

Knud

I generally divide the market into three segments: Consumer IoT, Industrial IoT—which primarily includes manufacturing and, to some extent, the energy sector, depending on how you categorize oil and mining—and Non-Industrial Enterprise, such as healthcare.

We no longer focus much on the consumer sector. Initially, we explored this area but realized that it operates with a completely different dynamic, and we are not as closely connected to consumer-related developments as we are in China and the USA. Our focus is strongly on the manufacturing sector, which holds the largest share in this space. However, there are also many exciting use cases in building management, the energy sector, and other areas.

That’s interesting because you just distinguished between Consumer IoT, Industrial IoT, and healthcare. Do you consider healthcare as a separate category?

Knud

Yes, for us, healthcare falls under the Non-Industrial Enterprise category.

That’s intriguing, as many of our users primarily focus on traditional production use cases but are increasingly linking them with building data and other domains. The boundaries are becoming blurred—it’s no longer just about production but about cross-functional applications.

Critical infrastructures like water or gas supply often have very different requirements. It’s challenging to clearly separate these topics because use cases are becoming increasingly cross-sectoral.
Or do you consider manufacturing primarily as an industry?

Knud

Both. Of course, there are overarching topics, but buildings operate under different logic than machines. For example, in building automation, protocols like BACnet are relevant.

The key players in this space are companies like Johnson Controls and Honeywell. On the factory floor, however, we talk about control systems from Siemens, Beckhoff, and others, as well as machines from the German mid-sized sector and beyond—they speak completely different languages. There are connections, of course, but I would argue that they are often marginal.

Tobias

The points of contact are also completely different. Classic building technology falls under smart building use cases, typically handled by building technicians or facility management organizations. In contrast, smart manufacturing involves professionals with a completely different educational background.

Perhaps I should clarify: When I talk about a use case, I mean a specific application such as automated doors in reception areas. Here, data is collected on when doors open or close to analyze heat exchange. A similar use case exists in production with industrial roller doors that open and close for different reasons. The use case is essentially the same; only the environment differs.

Knud

Yes, I understand that. While there are overlaps, I would classify them as relatively rare. A factory building is primarily seen just as a building—similar to any other type of facility. The biggest overlap, in my opinion, is in the energy sector. It’s about analyzing the total energy consumption of a building, identifying the largest consumers, and uncovering potential savings. A key aspect here is the holistic measurement of these factors. This creates a strong connection between the assets within a factory and the actual building management.

Exactly! I often think about how an existing use case can be applied to other areas. In our community, the question frequently arises: Is there someone who has already solved a similar problem—whether in another function or industry? That’s something I focus on a lot, hence my question.

Tobias

In market research, one of our main tasks is to categorize things in an understandable way. Use cases can be structured in many ways. Some define them based on the IoT devices used, such as smart metering or connected cars. Others classify them by functional aspects like monitoring, remote control, or optimization.

Another approach is to categorize them by application areas, such as quality assurance or maintenance. There are often overlaps—a classic example is predictive maintenance, which combines many of these elements.

Ultimately, each company has its own requirements and perspectives, making the categorization of use cases quite complex.

Yes, that’s an important point. When I collaborate with partners like Portainer or others, many technical topics are also seen as use cases—not just the traditional business applications. The question is: What exactly is a use case?
Technical topics such as device management or event-driven architectures—like those from Confluent—are often considered use cases. While they are technical solutions, they are closely tied to business requirements, as companies need to address these challenges.

Tobias

I would rather refer to them as challenges in this context. We operate in various industries—whether it’s automotive, smart manufacturing, the energy sector, or defense—and the challenges related to data extraction, processing, or forwarding are often very similar. And I agree with you: those with experience across different industries can benefit and adapt proven approaches—always taking industry-specific details into account. What works in retail, for example, can’t be directly applied to a production plant.

Exactly! A use case could be condition monitoring of vibration data or asset tracking of pallets. Behind it is always a challenge that needs to be addressed—ultimately, it’s a business case.

However, for me, technical topics such as device management, for example with Docker, are also valid use cases.

Knud

We have different perspectives on this. For me, device management and event-driven architectures are part of the tech stack—they support the implementation of a use case but are not use cases themselves.

Yes, it’s an interesting topic to discuss. What’s fascinating is that companies like Confluent explicitly label such technical solutions as use cases and list numerous technical applications on their websites.

Knud

I see the difference in the fact that a true use case is always tied to a business case—in other words, it can be calculated, and a tangible benefit can be derived from it.

A major challenge with IoT is that many necessary investments go into the underlying infrastructure without companies being able to link them directly to a concrete application. This often means that these investments don’t pay off immediately. If I implement an event-driven architecture, I haven’t made any money from it yet. The crucial point is where it delivers a specific value. A real use case only emerges when it is connected to an application that provides a clear benefit.

Tobias

U.S. hyperscalers market exactly that: they sell companies on the implementation of so-called use cases by first creating data highways to the cloud—often without a clear use case in mind. Their motto is: let’s collect data first, and the use case will come later. Meanwhile, they are already generating revenue.

One of the weaknesses in IoT project implementation in the past has been launching too many projects without a clear business case. Data was collected without having a concrete goal in sight.

Absolutely! For me, this technical use case does have a clear business case. Take Kafka or Confluent as an example—simply because it’s top of mind for me. It’s easy to calculate that using such technologies can shorten time-to-market. Scalable technologies also offer opportunities to reduce integration efforts and address similar challenges. The same applies to device management, where technical solutions often come with solid business cases. In that sense, I also consider such technical implementations as use cases.

Knud

That might be true for machine builders, but for companies like Volkswagen, which use such technologies in production, I’m not so sure.

Tobias

In practice, we often encounter the classic “never touch a running system” mindset, especially in the smart manufacturing sector. At the field level, update mechanisms are often still lacking.

When security vulnerabilities are discovered and urgent updates are needed, the absence of the right technology often means that employees have to manually update devices by walking through the plants with USB sticks. This process is time-consuming, error-prone, and results in high personnel costs.

Knud

Tobias, how is this handled with your customers? Are their budgets tied to specific business cases, or are they more general investment decisions aimed at modernizing edge infrastructure?

I often hear that it’s challenging to justify investments without clear use cases. I would have positioned you in a space where you don’t directly offer solutions for predictive maintenance or vision-based quality. You may enable these applications, but you don’t execute them yourself. How do your customers feel about this?

Tobias

The advantage is definitely there. However, as you rightly pointed out, it’s often difficult to convey this added value to financial controllers or decision-makers. We see this frequently with our strategic partners, who also struggle to sell additional aspects like hardware or device management. These elements are usually integrated on the fly.

The main driver for edge implementation is always a specific project or solution—whether it’s predictive maintenance or quality control. Within this framework, various products and infrastructures are then implemented. It’s rare for a dedicated business case to be created solely for device management; instead, it’s usually part of a broader solution.

Another exciting topic is managed services. Some of our partners work with technologies like HiveMQ, which cover this area. Companies such as Endress+Hauser, IFM, or Turck operate IoT platforms for their customers and manage thousands of devices in the field.

These manufacturers face the decision: should they handle device management themselves, or should they rely on managed services from a partner? Should they, for instance, collaborate with Portainer to make their device management more scalable?

From my perspective, the business case for this is relatively easy to calculate—such as measuring the time savings from eliminating manual processes like distributing updates via USB stick. However, the challenge lies in calculating the underlying scalable IT architecture, which makes it complex. But to me, this is already a business case.

Knud

As I mentioned, I see this trend more applicable to device manufacturers and machine builders rather than companies that use these technologies in their production. Volkswagen is just one example, but this applies to many manufacturing companies. If the technology is solely used for cost reduction and doesn’t directly generate revenue, it becomes a challenge.

However, Schaeffler, for instance, has also built its messaging infrastructure on NATS, and they must face the same question.

Knud

In a webinar two or three years ago, I heard that they strategically allocated an infrastructure budget. This budget wasn’t justified through individual business cases but was seen as a strategic initiative for modernization. Companies can apply for funding to implement their projects.

I suspect that many companies invest in certain areas simply because they see it as a necessity.

It’s an interesting discussion: where are such decisions made, and what ultimately qualifies as a concrete use case?
Let’s wrap up this topic for now—we can always continue the discussion offline. By the way, I’d love to hear what our listeners think about this.

Feel free to share your thoughts in the comments or reach out to me directly on LinkedIn—it’s always a fascinating topic.

Let’s move on to the latest trends in the IoT space. Which developments have the potential to become truly practical?

Tobias, in your view, what are the most important trends?

Tobias

There are many trends, but the question is whether they are truly new—after all, things don’t suddenly change on January 1, 2025.

A key trend is IT-OT convergence. IT and OT are coming closer together—both technologically and in terms of workflows. This trend is particularly evident in the manufacturing sector. While traditional IT and OT departments still exist today, hybrid units like production IT are increasingly emerging, combining both IT and production expertise.

This means that traditional structures are gradually dissolving, and teams with mixed competencies are working more closely together.

Tobias

Exactly! This might be a slightly different answer than the one typically expected—where the focus is solely on AI.

Knud

IT-OT convergence is a topic we’ve studied extensively. We analyzed 27 aspects within this field, including how teams are evolving.

Two main approaches have emerged: Some companies choose to establish dedicated units where IT and OT experts work together and report as a single entity. While this approach is not yet widespread, many see it as the way forward. A notable example is Toyota, with its Production Engineering unit present in both Japan and Europe. In some cases, these teams are further complemented by business value specialists or data scientists. Other companies, however, maintain their separate IT and OT departments but frequently assemble interdisciplinary teams for specific projects. This flexible model is already well-established in many organizations.
Coming back to IoT trends—besides IT-OT convergence, AI remains a major focus, particularly generative AI. In IoT and manufacturing, predictive maintenance and machine vision for quality control continue to be two of the most important application areas.

Predictive maintenance initially wasn’t true AI in many cases, but now we are seeing some exciting implementations where AI is genuinely being applied.
New trends like industrial co-pilots and generative AI are gaining traction. However, there is still a lot of show—whether it’s Siemens using generative AI for code generation or other providers using it for guided maintenance. What’s truly exciting is when AI is used to analyze production data in natural language, such as: “Show me the production data from the last time employee XY worked under weather conditions XY.”
Another important topic is sustainability, although the trend has been declining. For me, sustainability is divided into two major areas: first, the carbon footprint and everything associated with it, and second, energy savings, which are related but have a different focus.
In Europe, sustainability remains a key topic, often driven by idealistic goals. However, in the US, interest has significantly declined—especially since 2021/22. Unfortunately, our analyses show that the topic is losing momentum. Nevertheless, there have been some successful implementations in the field of sustainability.

Tobias

I’d like to jump in on the topic of AI. One argument for AI as a sustainable trend is that it’s being applied across industries—not just in smart manufacturing or logistics, but also in retail.

A great example is self-checkout systems, where AI-powered cameras can detect whether customers are scanning the correct products with the right weight—or attempting fraud by covering sensors with their hands.
Another powerful use case can be found in the MedTech sector. A particularly touching example is early sepsis detection in premature babies. Every second counts when it comes to administering antibiotics in time. AI analyzes specific vital signs in real time and alerts doctors to anomalies. AI models, running directly in hospitals, detect abnormalities based on critical metrics, enabling rapid intervention and immediate alerts to healthcare professionals. This demonstrates how AI implementation is progressing across multiple sectors.

Knud

Let’s get back to IT-OT convergence. What does it actually mean? At its core, it’s about bringing IT and OT together. Often, OT adopts technologies and methodologies from IT, while the reverse is less common.

The question is: Is AI an IT technology? If you define it that way, then yes—there’s a clear convergence. However, I see AI and IT-OT convergence as two separate topics, even though AI is now being integrated into various areas.

In the coming years, AI will be ubiquitous. Eventually, we won’t even need to talk about AI use cases because almost every solution will incorporate some form of AI. Of course, AI will play a role in IT-OT convergence in different ways.
You just mentioned edge—shouldn’t that be considered a trend in its own right? We now have significantly more processing power available on-site, with a completely new software stack compared to the past. The deployment of GPUs and specialized AI processors is making edge computing more powerful than ever.
There are exciting prospects, such as the possibility of running local LLMs (Large Language Models) directly at the edge instead of relying on cloud services. However, much of this is still in the future.

Tobias

Exactly. Many customers carefully consider their strategies and don’t just follow the trends that hardware manufacturers push—such as integrating large computing units into every control cabinet just to maximize available resources. Instead, a targeted approach is taken to utilize edge computing, partly for cost-related reasons.
I see this as a continuing trend – but more targeted and geared towards several use cases.

Absolutely, that makes sense. Considering the growing number of connected devices—Knud, you recently published a study showing that the number of IoT devices has grown by over 13% to 18.8 billion. The development is ongoing, which makes it crucial to create scalable solutions for the future.

Knud

Yes, we live in an increasingly connected world, and this number will continue to rise. However, it largely depends on how you define it. Based on our definition, there are currently around 19 billion devices.

Another exciting IT-OT trend is Industrial Data Ops, which is steadily gaining traction. You could consider it a part of IT-OT convergence—or even the new IoT platform.

Industrial Data Ops provides a native data platform for the edge, connecting various systems, contextualizing data, and offering connectors for the cloud and other data lakes. Technologies like MQTT play a crucial role in this space.
There are now several promising startups in this area, and large enterprises are also showing increasing interest in the technology. I see Industrial Data Ops as the next evolutionary step of IoT platforms—only now, the term “IoT platform” seems outdated.

That’s an interesting topic! Perhaps we should explore it in a dedicated discussion: Is Industrial Data Ops the new IoT platform—and is it a real trend?

Another topic I’d like to bring up before we wrap up is change management, which remains a significant challenge for many companies.
Many businesses are facing the question of how to enable their organization for IoT. New departments and roles are emerging, such as Digital Agents, or simply responsibilities aimed at identifying and driving use cases internally.
I propose the thesis that one of the biggest challenges is not just implementing IoT but identifying the right use case for the company and consolidating internal knowledge.
What’s your take on that? Is change management the biggest challenge in identifying IoT use cases—and why?

Tobias

One crucial factor is having prior exposure to a use case or at least a fundamental understanding of it to implement it successfully.

Often, we see that many employees—especially in Europe—have been with a company or in an industry for a long time. While job titles and responsibilities evolve, their knowledge often remains at the level it was when they graduated 15 or 20 years ago.

Of course, lifelong learning is an important aspect, but I believe it is still not sufficiently practiced in many companies. In large corporations, fresh insights often come from external consultants, who bring perspectives from other industries and support internal training efforts.
However, looking beyond one’s own industry and drawing inspiration from other sectors is crucial. Platforms like IoT Use Case provide exactly this opportunity to explore new use cases independently of external consultants. This is not only important for engineers but also for management-level decision-makers.

Knud, what’s your take on this?

Knud

We conduct various types of market research—one major component is surveys, where we ask companies: If you could start the project over again, what would you do differently?

The most common response is: Involve stakeholders earlier, take change management seriously from the beginning, and engage users sooner. This pattern emerges across nearly all surveys. While companies are aware of it beforehand, it continues to be underestimated.
You mentioned that the biggest challenge lies in identifying use cases.

From my experience, companies don’t necessarily lack ideas or an understanding of what they could do. The real challenge is prioritization. The key question is: Which use case should we implement? Everyone is aware of their 20 challenges and has numerous ideas to address them. The issue is not a lack of ideas but determining which use case delivers the greatest value and how to make that decision.
A great example is companies where strategic decisions aren’t made by a CEO simply declaring, “We’re investing 100 million in generative AI—go do something.”

Instead, some companies have cultivated a culture of local innovation over many years.
A large German chemical company, for example, has established a structure where employees can contribute ideas locally and test them in small projects. They have a dedicated budget for pilot projects, which can then be rolled out across different locations. This company recently rolled out an AI assistant globally – originally developed in a single factory. Such structures are a crucial part of the solution and a key to success.

Tobias

However, this is a luxury that primarily large companies can afford. In a big corporation with a dedicated IT department, such software projects are easier to implement. In SMEs, on the other hand, this is a greater challenge, as a small number of IT managers often also have to ensure day-to-day operations.
Another important aspect is the complexity of IoT projects. We’ve already discussed how they are becoming increasingly demanding. There is definitely potential to manage this complexity—the key question is how to do it effectively.
While internal teams often have the right ideas, interdisciplinary collaboration plays a crucial role, and there is still significant untapped potential here.

Absolutely! Several aspects come to mind. A key challenge is identifying the right use cases.
Many manufacturers struggle to understand how their machines and devices are actually used in the field. From my network, I often hear that machine builders or companies with devices and machines that have been in the field for years often have little insight into how they are actually being operated. The key question then is: What value can we offer, and what is the customer actually willing to pay for? It’s all about finding a use case that creates real value for both parties. This challenge is not limited to smaller companies but is also faced by large corporations.

I have a specific company in mind, though I won’t mention it publicly. They have been dealing with IoT for 15 years but are still trying to determine which use cases truly have potential for the future. For example, take cleaning machines that have been in use for years. The real use cases—such as utilizing data from these machines for refilling cleaning fluid or predictive maintenance—are often still not clearly defined.

I think manufacturers face the major challenge of accurately identifying these use cases to answer the central question: What is the end customer willing to pay for, and how can digital services be effectively expanded?

Tobias

Yes, but at the same time, suppliers and product manufacturers naturally want to position their own products. However, simply purchasing a product is far from being a solution. A true solution often arises from combining different approaches—whether it’s a “one-size-fits-all” or a “best-of-breed” approach, where products from different or even the same manufacturers need to be integrated. These concepts must be evaluated together with the customer to determine the actual needs.

Exactly! This creates the challenge of not only prioritizing this knowledge but also making it accessible for implementation. A major issue is that use cases successfully implemented in one plant—say, in Herzogenaurach—are not automatically known in plants in the US or China, even within the same company.

The knowledge about successfully realized solutions is often not readily available. The challenge, therefore, is less about identifying use cases and more about ensuring standardized implementation based on previously gathered experiences, pitfalls, and best practices.

Tobias

Another key topic is centralizing expertise. Especially in large enterprises with multiple locations, establishing a Center of Excellence can be beneficial. Such centers can consolidate the collected knowledge and make it available for scalable rollouts.

Knud

That makes sense. About a year and a half ago, we conducted a survey with machine builders and various equipment manufacturers. We asked them: What is the most successful use case for your customers—what works best for them?
We provided around 20 options, including measuring the energy efficiency of machines and monitoring inventory levels. If I recall correctly, the most frequently mentioned feedback was that the greatest value lies in helping customers use their machines as efficiently as possible.
What exactly “efficiency” means, however, is something that each industry needs to define for itself. Our study showed that this aspect provides the most value to customers—based on the limited data available to the participants at the time.

Yes, exactly. In the podcast, I often deal with these two perspectives. I recently made an episode with CLAAS, the well-known agricultural company. They run CLAAS connecta platform hosting numerous use cases— some well-known, others not yet fully explored. The focus is on how data can be used even more effectively in the future.
That’s one side of the story. The other side is the practical implementation of these use cases in operations.

Do you have any final best practices or insights on successful implementation? Tobias, looking at you—what do you think are the most critical factors for successfully implementing a use case?

Tobias

From my perspective, the first thing needed is a clear and concrete concept. I agree with Knud – a business case is important to get a project off to a good start.
Simply diving in and experimenting can be useful in a lab environment, but in real-world scenarios, everything should start with a well-thought-out business case, which can then be broken down into individual use cases.

Strategic thinking is crucial—not just whether something works once, but what the long-term implications of a decision are. When selecting specific products, one should be aware of what this means in five or ten years. Will it create dependencies? These aspects must be considered. It’s not just about choosing the cheapest product from a vendor who might not even exist a decade from now. Relying on a specific American cloud provider, for example, could result in vendor lock-in, which might become problematic in the long run. All these factors should be part of the decision-making process, and fortunately, there are increasingly more alternatives worth considering.

Maybe one final question at this point—greetings to Boris from KUNBUS, whom I spoke with earlier today about this episode. He raised an interesting point: Looking ten years ahead, how will IoT solutions change for users? How will their jobs evolve? What do you think?

Knud

I believe that generative AI will play a major role in user interfaces. We will interact much more frequently via voice commands rather than manually configuring dashboards. Instead, we might simply use voice input and even receive automated suggestions for the next logical step.
I think the user interface in the IoT world—and in technology and IT in general—will evolve significantly. Furthermore, I expect that the term “IoT” will fade into the background, as connected solutions will become a given. Hopefully, ongoing efforts by many companies will make access to data easier. Today, data is increasingly being centralized in a data lake, which many saw as an unviable strategy just a few years ago. However, most now agree that cloud-based data lakes represent the future.
An essential factor will be a well-structured access management system that ensures everyone gets access to the exact data they need—quickly and seamlessly.
We’ve long envisioned instant access to all relevant real-time data. Solutions like SAP’s ERP systems have provided initial steps, but I hope future data access will be more seamless and intuitive.
User interfaces will continue to evolve, manual tasks will be reduced, and many processes will be automated by AI-powered systems—think of agentic AI. Working alongside AI-driven assistants could lead to a reduction in personnel needs while simultaneously boosting efficiency.

I share the view that by 2035, we will hopefully achieve significantly improved data availability. A centralized data foundation would allow for quicker and easier implementation of use cases. But this still requires a lot of groundwork within the industry.

Knud

However, you asked for a realistic outlook, not a wishful scenario. Looking back over the past ten years, I can see that many of the challenges we faced back then still haven’t been fully resolved today. As long as people remain involved in processes, we will continue to face typical challenges within information and communication chains. These challenges will continue to accompany us in the future.

Tobias

I believe that even in ten years, there will still be significant potential for optimization, especially in process management, where considerable efficiency improvements are still possible.

One aspect I strongly support is the increasing use of data—not just for analysis in data lakes but also for the gradual automation of decision-making. This automation will go beyond what we see today, although there will ideally still be critical points where humans can intervene and make decisions. Many see AI as a short-term trend that could fade within the next ten years—similar to patterns seen in the Gartner Hype Cycle. However, I don’t believe this will be the case. AI is often viewed solely through the lens of ChatGPT since 2022, but the concept has been around since the 1950s. AI has already gone through periods of disillusionment in the 1970s and 1980s. In the coming years, however, I see AI progressing towards a plateau of productivity, which we will eventually reach.

Knud

One point that I’m sure Boris from KUNBUS would agree with—and which I also see as likely—is the increasing standardization of hardware. In the future, we will see a greater reliance on standardized solutions, such as industrial versions of the Raspberry Pi or similar platforms.
I believe that hardware platforms, which are still highly specialized in today’s production environments, will increasingly become interchangeable standard solutions. A key term here is the virtual PLC (programmable logic controller),which is currently at the forefront of development and something we’ve been closely monitoring.
I think this could play an important role in the future, as it would help reduce dependency on embedded hardware experts for specialized systems and operating systems. Instead, companies could rely more on standardized hardware solutions.

Tobias

Hardware will indeed become more interchangeable—much to the delight of customers. But when we look at software PLCs, another very exciting topic, there are still challenges that need to be overcome, particularly in the area of safety.

Knud

As far as I understand, some of these challenges have recently been addressed—both by Siemens and CODESYS.

Tobias

That definitely needs to be examined in more detail, but there’s certainly a lot of potential in this area.

That’s a great outlook for the next ten years and a perfect closing statement for today. I want to thank both of you for being here today. I found this conversation incredibly insightful and could easily spend another hour discussing it with you. Maybe we should do a special episode on this topic in the future. I’ll link all relevant sources, reports, data, and everything we’ve discussed in the show notes, along with your contact details.

If you, our listeners, have any thoughts on today’s episode, feel free to leave a comment or reach out to Knud and Tobias. We’re looking forward to your feedback.
From my side, a big thank you—and now I’ll leave the final words to you. Who’d like to go first?

Knud

All I can say is—if you’ve listened this far and find the topic interesting, feel free to stop by IoT Analytics if you’re ever in Hamburg. Otherwise, you can find me at trade fairs like the Hannover Messe or SPS. Just ping me, say hello, and let’s connect.

I always look forward to exchanging ideas—it’s the essence of my business, and I’m happy to share my knowledge wherever I can. So, here’s my call to action: Connect with me on LinkedIn.

Tobias

Yes, the same goes for us. We are also present at all the major events and exhibitions.
A big thank you to Madeleine for organizing this—it’s not every day you get to record a podcast right on the Reeperbahn, so thanks for making that happen! I really enjoyed the setup, as well as the independence with which we could explore the market.

Knud

I’d say we’ll see each other at the next IoT Use Case event or one of the upcoming trade fairs.

Yes, I was just about to mention it— be sure to stop by! The next IoT Use Case event is listed on all major exhibition calendars. We’ll be there, and I believe you both will as well. See you there then!
Thank you all, and have a great rest of your week!