Transforming Asia Pacific's Digital Future: IBM's Vision for Enterprise AI with Hans Dekkers
Fresh out of the studio, Hans Dekkers, General Manager of IBM Asia Pacific, joins us to explore how enterprise AI is reshaping business across the region. He shares his journey with IBM after business school, reflecting on the evolution of personal computers to AI today. Hans explains IBM's unique approach combining hybrid cloud infrastructure with AI for Enterprise, emphasizing how their granite models and data fabric enable businesses and governments to maintain control over their data while scaling AI capabilities. He highlights customer stories from Indonesian telecoms company to internal IBM transformations, showcasing how companies are re-engineering everything from HR to supply chains using domain-specific AI models. Addressing the challenges of AI implementation, he emphasizes the importance of foundational infrastructure and governance, while advocating for smaller, cost-effective models over GPU-heavy approaches. Closing the conversation, Hans shares his vision for IBM's growing presence in Asia as the key to enterprise AI success.
"At IBM, we really work on two emerging technologies: hybrid cloud and AI for enterprise. These two are deeply connected. Hybrid cloud for us means that regardless of where the data sits whether the compute is on-premise, off-premise, or across multiple clouds. We believe the client should have the control and flexibility to choose where to run and place their data. If you look at the facts, a very high percentage of client data is still on-premise. It hasn't moved to the cloud for obvious reasons. So, how can you scale AI if you don’t have proper access to that data? AI is all about the data. That’s why we believe in a strategy that redefines and rethinks everything. We call it the Great Technology Reset." - Hans Dekkers
Profile: Hans Dekkers, General Manager IBM Asia Pacific and China (LinkedIn, IBM)
Here is the edited transcript of our conversation:
Bernard Leong: Welcome to Analyse Asia, the premier podcast dedicated to dissecting the pulse of business, technology and media in Asia. I'm Bernard Leong, when we talk about enterprise AI, IBM is definitely at the forefront. So with me today, Hans Dekkers, General Manager IBM Asia Pacific, I want to understand what is the impact of AI in the Asia Pacific? Thanks to the team from IBM and I get to host this show with your video production team. So welcome to the show, Hans.
Hans Dekkers: Thank you. It's so good to see you. You're becoming a superstar on the internet, so the privilege is absolutely ours.
Bernard Leong: I'm sure my guest is actually the reason why I'm here. So while doing research for this interview, I was struck by your global journey by steering the world's leading tech firms in the Asia Pacific. I wanted to start with your career journey. What drew you to IBM immediately from your business school experience?
Hans Dekkers: First of all, I was one of those kids that genuinely is deeply interested in technology. So as a kid programmed, the first games, used a Commodore 64. Maybe it's before your time.
Bernard Leong: I remember the Commodore 64, the IBM PC and IBM Pentium.
Hans Dekkers: IBM PC, the 286, the 386, the first Intel processors that came. But I'm also very much into disassembling cars and radios. We used to pull everything apart we're trying to put it back together.
Bernard Leong: That is not like these days where everything's integrated together you don't know what's inside.
Hans Dekkers: Like try to fix a Dyson, right? That's right and not easy. So very much technology educated, in all the technology from network infrastructure design, chip design, programming. Very deep technical. Did my business studies at my own companies, then really stopped in saying what's next? Where do you as a young professional start your career? There are many places I was privileged to start at big banks, at big trading firms, many technology companies. But in the end, when I was still young. I asked, "Who's the biggest technology company out there? Who is the one that actually impacts the outcome of society?" When you really take that aperture, they're only very, very few companies. IBM was the one for me at least that came on top. So I started an internship as a trainee at IBM about 17 years ago. That's how the journey started.
Bernard Leong: So through IBM, you led different geographies from Europe to Asia. What led you to Singapore can you talk about your current role for IBM Asia Pacific?
Hans Dekkers: So first, I'm one of those global kids, so I lived a large part of my life in Asia, a large part in Europe and in Northern America. Always in IBM, I took on different roles. I've been from systems to software to coverage to strategy, always with clients. I love talking to the client, always at the heart of where you can make the most impact. I always loved to have the privilege to leading big teams. Really get the best of that IBM team in service of your client or your partner to get to a real outcome. So many different roles, many different geographies coming back to Singapore because it's the second time we as a family live in Singapore. About 10 years ago, we were here as well. It's just a beautiful region. It's got the deepest cultures, the longest history. It's got 60+% of GDP growth. It is the geography that is always on the move. China, South Korea, you've got Australia, New Zealand, you've got India. Always something's happening, any dimension. So it's great to be back.
Bernard Leong: So looking back, given the 17 years with IBM, what are the pivotal lessons you can share about your career journey to a younger audience out there?
Hans Dekkers: I think we always say, finding the solution actually is not that difficult. It's finding the right problem.
Bernard Leong: Interesting.
Hans Dekkers: In IBM, there's always, if you find the right problem, you supplement it with data with science. If you find the right problem, then you can work together with your client in actually addressing it. As a technology leader, as a technology company, we're privileged because in some topics we can see that future, maybe our client cannot. Then to help a client from point A to point B is incredibly rewarding.
Bernard Leong: So you find that whole customer journey, the ROI that you achieve, that is the most rewarding?
Hans Dekkers: Yes. To do something that the client or the government or whoever you are working with seem to be impossible, to do something that is truly meaningful for them for society. It's really what keeps me going.
Bernard Leong: So we are gonna come to the main subject of today. I want to talk about IBM in the Asia Pacific achieving AI ROI. Even when I was working for one of the cloud companies, I think a lot of people don't look at the underlying numbers and one of the highest revenue drivers for AI is in IBM.
Hans Dekkers: Right.
Bernard Leong: So I want to explore the intersection of AI enterprise growth with IBM's current strategy. So given that the Asia Pacific region is now experiencing so much rapid digital transformation, what is the scale of the AI market opportunity here in this region from your perspective?
Hans Dekkers: Unparalleled. I don't think we can even quantify how big that impact that opportunity will be. We have to recognize, maybe we can discuss a couple of foundational elements to "what is AI? what is it based on?". In the end, how do you scale it? Because it's all about scaling that capability. If you keep it small, what's the in the end outcome?
In IBM, we really work on two emerging technologies. One is what we call hybrid cloud. The other one is AI for Enterprise. These two are deeply connected. Hybrid cloud for us is that regardless of where the data sits, regardless of where the compute is, on premise, off premise, multiple clouds, we believe that the client should have the control flexibility on where to run place their data. Now, if you just look at some facts that a very high percentage of client data is still on premise. It hasn't moved to cloud for obvious reasons. Then how can you scale AI if you don't have proper access to that data? Because AI, it's all about the data. So we believe in a strategy first that redefines, rethinks, we call it the great technology reset. Rethinks that foundational capability of hybrid. Can we access tap into, secure, govern the full foundational landscape of infrastructure data, number one. If you are able to start there, you can start scaling AI use cases. If you don't have that in place, with the right proper governance, you'll struggle as an enterprise. So first step where we help clients is this, we create a horizontal platform, data platform across all of their estate.
Bernard Leong: So it's like chaining all the different data layers for the client, whether it's on-premises or on the cloud, but then it just makes sure that they have the correct security governance so that the AI can access do work with whatever data they have.
Hans Dekkers: Right. We do it by not moving the data. In many cases, we leave the data where it is, but we create that horizontally connected platform. Underneath you could do the same with your infrastructure because many of our clients started on a journey to cloud with no end. So they're using the hyper-scalers, which is fantastic. It's really easy to start with. If they grow bigger, maybe it becomes less cost effective or geopolitical situation changes. You want to move it from a hyperscaler to on-premises. If you design your infrastructure architectures correctly, you're able to move fairly easily on-premises, off-premises from X86 to maybe mainframes or from mainframes to X86 vice versa. You're very flexible. So both infrastructure layer, data layer, we believe that it needs to be horizontally connected.
Then we get to AI. Now, when you look at AI, we believe in two fundamental things. First, your data needs to remain your data. It's no one else's data. It's your data, which is very important if you're an enterprise client or the government of Singapore. Your data is your data. There's nothing IBM or any other company for that matter, should have to do with your data. It's your data. It's your intellectual property. Second one is that we believe that the AI models that we build are small nimble. They need to be cost effective precise. They should can be owned governed by you as an end client, which is very important. Then there's a third piece that we believe it needs to be open. So the models we govern need to be open enterprise, grade ready, domain specific so you can get the most value from it.
Bernard Leong: So my understanding, given that you talk about the two sites first, is the data infrastructure. Then there is a second part, which is the AI. Of all the AI stuff, I know the most groundbreaking is definitely Watson. The first AI to beating Garry Kasparov in International Chess. This is long before the days of AI playing Go and you know, also beating Jeopardy. Can you tell me about how IBM thinks about AI as a product or maybe service for the customers?
Hans Dekkers: Right. I'll get to that. You started at chess, which, the chess computer in that time, it was the first time that a computer took up the challenge to the human brain, Kasparov.
Bernard Leong: Yes. Kasparov.
Hans Dekkers: Kasparov lost that match.
Bernard Leong: The Deep Blue match.
Hans Dekkers: The Deep Blue match. Now chess at that time, no. But now I would say it's a very simple game because it's a game that is provisioned on finite set of moves.
Bernard Leong: That's right.
Hans Dekkers: Now we have chess computers that can easily calculate, you did this, therefore I can do this. At the time it was groundbreaking. So it was really the starting point of a computer against AI. It's the very first ingredient where we saw what could be possible. Then Jeopardy came.
Bernard Leong: Yes. That was harder.
Hans Dekkers: Much harder because it was based on natural language. But not in a way that our AI models work today. So it was more based on machine learning. It was still based on logic.
Today, it's a much more math based prediction of what's the next letter, what's the next word, in an AI model, which brings us now to the conversation of how IBM looks at AI. We fundamentally believe in three big components of AI. One is the data, we call this Watsonx data. It's that data platform across multiple infrastructures. It's your data lake. The starting point of this is what we call Watsonx AI, which is a studio, where you can bring together multitude of models. It's not one model. It's a multitude of layering of models where you combine it into your model as an enterprise or as a government. So you can create in our studio your model. The third piece as important is a governance piece. Watsonx governance. Is the model biased? Does it use the right language? Is it aligned to how you communicate to your clients? So you can do it the same way as almost you teach your children. Don't use this language. If you want to be on time, be on time. So it's almost governing the output of the model you can tweak it. So it's a controlling function which is very important.
Bernard Leong: I remember because while doing research, I also know that you have a data fabric layer that actually allows you to go even multi-cloud architecture.
Hans Dekkers: Correct. I think we're the only company that goes across a multitude of these clouds.
Bernard Leong: I was advising a retail client in Dubai on that. I worked out that you guys are the only ones with the data fabric, but maybe given the opportunity. So you explained the data structure. You talk about the AI. How does IBM uniquely position to capture that opportunity in Asia Pacific?
Hans Dekkers: So it's not only Asia Pacific, but I would say IBM wide. We believe that our CEO talks about this a lot: the cost of AI is still too high. It's about a hundred times too high. It has to come down. As with every technological innovation, if it's cheaper, more people can use it. Once it becomes even cheaper, it will become natural to the way we operate. One of the main reasons to make it cheaper is to use smaller domain specific models. Small models are very accurate, we can train them really well. We don't need massive amounts of GPUs to train these smaller models. You need less because they're smaller. We believe in these smaller models, there will be a place for big models. About 10 to 12 of them, a dozen of them will exist in the world, but the rest all will be small.
Bernard Leong: Can you give an example of a smaller model?
Hans Dekkers: So imagine you are a bank, imagine you are on a trading floor. You don't want your model to be good trained in Russian poetry. That's probably true. So you want it to be really good at running financial analysis on maybe the market or in certain stocks based on your history, based on all the data available in your enterprise. We help our clients with those domain specific models.
Bernard Leong: So specifically, what are the top challenges for enterprise when it's in this region, in the Asia Pacific region? I'm sure you come across a lot of different kind of clients. Again, I guess maybe when you talk to them, there must be some generic set of challenges they face. What are the common ones?
Hans Dekkers: I think many of them are leaning in very heavily to AI have leaned in into, for example, the hyperscalers a lot. I love Asia because they start, they don't talk talk talk. They start, they get going. While saying that, a lot of where we are today needs a lot of foundation, foundational re-engineering or optimization. Many enterprises haven't gone there yet, so they want to jump onto AI. But if you're not in a hybrid by design environment, it becomes very difficult to make best use of AI capabilities that are out there. First step I would say is to truly gauge where am I in my technical health? Where am I where do I need to improve quickly to make best use of scaling AI use cases? This is where I think Asia ASEAN clients can go very fast.
Bernard Leong: Then the question then is how do then the clients think about the ROI for AI?
Hans Dekkers: I think today they're still experimenting. They don't have a good idea yet. There are many numbers out there of productivity improvements. We've got a lot of our own experience, but in the end, when you project a use case that we have done with our clients onto their business, then I think if we really go through help the client see it, I think that journey becomes very meaningful.
Bernard Leong: Are there, like, without mentioning specific clients yet, because I'm gonna come to that later, but are there very specific kind of metrics? Like is it really in terms of productivity savings or is it something to do with say, revenue growth? That really interest the clients more.
Hans Dekkers: I would say in the end it's about redefining how clients operate, how they process. So it's business growth. There's a lot of productivity gain, which is the low hanging fruit. There's optimization on how you communicate with client. There's optimization on HR. There's optimization in your supply chains. These are all productivity improvements, which are great will make you better in serving your end client. But I believe there's a huge opportunity to re-engineer the business model of the enterprise themselves.
Bernard Leong: So you definitely have done, many clients, are there really like an actual use case that you're very proud of it is a customer story that you can share?
Hans Dekkers: I always like to start with what we have done as a company because we are one big enterprise. So if you look at how the HR teams re-engineered, how we run HR within IBM. If you look at the ROI we can provide you those details. It's unbelievable. Our supply chain, how we re-engineered it. We create a lot of chips. We create a lot of machines, a lot of hardware, completely re-engineered the supply chain. If you look at the way we do forecasting the way we do accurate reporting, all of this has been completely optimized with AI. If you look at our consulting teams on how they optimize what they do for end clients, completely fueled with AI capabilities. It is a great shift.
Bernard Leong: So what's the one thing that you know about IBM's focus on AI that very few people do?
Hans Dekkers: Excellent question. So IBM creates its own models. We call it IBM granite, the family of granite models. Which actually also aligns with what you mentioned earlier about smaller models the right expertise, because that is also the rationale of actually using the company's IP.
Hans Dekkers: A hundred percent.
Bernard Leong: So maybe now, given that you know, in Asia it's such a culturally diverse with so many different regulatory, how do you see the landscape for IBM's AI strategy to actually touch different markets? I think Singapore, we are lucky. We have all the AI policies done on there, but let's say we go into, say a country like Indonesia, where still at a formulation stage, but maybe even Japan Korea will have a good AI policy. How does IBM navigate it?
Hans Dekkers: I think in any maturity, depending on how you would create that maturity, I think IBM can be a great player. Everyone that wants to take that next step in technology actually can find a very good partner in IBM, including Singapore. Because the world outside US is changing so fast, geopolitical, it's changing. If you look at protectionism of data, for example, it's changing very fast. You got data laws that are being passed. All of this has a huge impact onto how to run your technology estate how do you get the most benefit from it. You have vendors that increase price dramatically. If you're stuck with them, that's a problem because it will consume all of your budget. If you're not flexible enough in making your own decisions, something COVID really taught us, then you're in a deficit. So it's the companies that built this agility of change make sure that their technology landscapes are ready for that change that will reap the most benefits. In Asia, depending on how you grade maturity. Again, I think every country, every enterprise can be on that journey. Some are more mature than others. But I believe anyone can benefit from someone in Indonesian Surabaya to somewhere in Pune in India. It doesn't matter where you are physically. I think the benefits of what we can do together is unparalleled.
Bernard Leong: So, Hans I want to specifically go into customer use cases. Can you share like any interesting customer stories where the IBM AI has actually helped customers in this region make an impact?
Hans Dekkers: I mean, many. I would say the most recent one, recent, is a collaboration, a true partnership we went into with a telecom in Indonesia.
Bernard Leong: Okay.
Hans Dekkers: Telecom is basically it's got a huge infrastructure, a huge infrastructure provider to government, to enterprises, small, medium businesses. They're gonna embrace IBM's watsonx platform for all use cases, which, if you look at Indonesia as one of the bigger countries makes total sense. Because your data is your data. You can control it, you can govern it. We can apply it to a multitude of use cases. So it's a beautiful example of where, in this case, a partner us are completely collaborating to bring AI to a country, a country of Indonesia in this case.
Bernard Leong: Nice. So like now, this year, a lot of people say this is going to be the rise of AI agents. What kind of opportunities do you foresee for these in the Asia Pacific region?
Hans Dekkers: I mean, GenAI agents, I think it's gonna yet again transform how we experience AI. We recently done a great POC where we had multitude of agents. HR, finance, legal, business functions basically debate with each other. We were following the debate. So we threw in a hypothesis, I want to create a presentation on something. They were debating each other. You could follow the flow by which they were debating they were constructing a narrative. Which gave us insights that we would not come up with. So this continuous evolvement of how capable agents are, what they could do is gonna change how we operate.
Bernard Leong: So for, I mean, given Asia Pacific is such a dynamic growing region, let's say, for companies that are still very early stage in their AI journey, what would be your advice for them to think about implementing AI successfully sustainably?
Hans Dekkers: So this is gonna sound difficult. It's not. Decide on infrastructure, plumb your data platform. Make sure you have the right tool set. Make sure you own the data, you own the model, then govern it correctly, then plug in the use cases that you want run.
Bernard Leong: So what kind of principles that actually guides IBM in terms of when they determine whether a new AI solution actually add value to its clients at enterprise scale?
Hans Dekkers: I mean, we do it together with our clients. So in many cases, our client comes with a problem, this is a problem I have, or this is a situation that I don't really have a solution for. Then we work together. We've got our client engineering teams. So we basically unpick that problem using this technology. That's one way. The other way is that we look at a company we say. This is the way you've always operated. What happens if we plug this technology onto how you always operate it? Could it look very different? There, it's really interesting. As soon as the client allows us to have that conversation, a whole new world of business models opens up. This will continue to evolve. So it's finding that connect between IBM the client on the right problem. Then the solutions will emerge.
Bernard Leong: So what is the one insight you wish more people would ask you about implementing AI that they don't usually ask you?
Hans Dekkers: I think the notion is that today everyone talks about AI equals GPU. GPU is important, but what I call inferencing is much more important. So creating that initial model, that initial philosophy, fairly simple. Then how do you train it? If you have a 3-year-old kid, it's great, but a 3-year-old kid cannot solve deep enterprise problems. That is done through continuous learning of helping, in this case a kid, progress into that enterprise state. This is where I hope clients basically work with us much more deeply. They think AI needs GPUs. I need to pump in all my data then I have an AI model. I can run a new business model.
Bernard Leong: Right. It's basically what I call the customer collaboration value, to sort of get the customers to work together to produce the solution that actually fits them, rather than thinking about just training the models. Am I, did I get that correct?
Hans Dekkers: It is directionally absolutely correct. How do we re-pivot to what's the outcome a client wants how do we get there fastest? Often the real work is in the training the tuning. It's not in the initial creation.
Bernard Leong: You think that inference is going to be much more important in the next couple of years?
Hans Dekkers: I think it will increase in importance. I also know that you'll need a very differentiated set of hardware to be good at inferencing. The GPU is fantastic, but it's really good at what it does today. It may not be optimized for tuning. So you'll see that there are what we call APUs, AI processing units, they're much better at inferencing.
Bernard Leong: So my traditional closing question, what does great look like for IBM in Asia Pacific in the next few years?
Hans Dekkers: I think it's reestablishing growing presence in doing technology in the right way the right way is a perception, but I really, together with all IBMers across the region will want to reestablish IBM for what it is today, not what it may have been 20 years ago. We're such a growing technology force. We've got such a differentiated view that I hope, I expect, I believe that in the years from now, we will be a much bigger technology company in Asia.
Bernard Leong: So, Hans, many thanks for coming on the show share with me the insights on what IBM is doing specifically with AI data in the Asia Pacific region. I have two small quick closing questions. First one, anything that have inspired you recently?
Hans Dekkers: Like what? A book or a movie, TV or even could be something inspiring, I think depending on the topic, there's so many inspirations. I like reading books. But also series, I don't have too much time, but once I do. You got great series now. I don't know if you watch The Last Of Us, it is pretty creepy. But it's fun to watch friends neighbours. Funny but very US centric. I would say. But I also like to go into history. So some of the history movies. Also on the region there recently was one on Korea, South Korea on the coup that happened. So I like things that are connected to deep stories. On books. I recently read a while ago "The Geek Way" by Andrew McAfee.
Bernard Leong: I haven't read that book yet.
Hans Dekkers: You should absolutely read it because it talks about some of the business fundamentals that I think are really important for future leaders it talks about radical candor, openness and scientific thinking. It's a great book to have your mind framed correctly.
Bernard Leong: I should pick up that book. But my last question, how do my audience find you?
Hans Dekkers: Find me in LinkedIn. Find me by coming to IBM, find me online. Very accessible. So if you have questions, if you have remarks. Please do reach out and we'll be in touch.
Bernard Leong: You can definitely find us on Spotify, YouTube, all the other channels, then subscribe to us. We just hit our hundred K subscribers. So many thanks for the support Hans. It was a great conversation. Many thanks for hosting me here, I look forward to speak to you soon.
Hans Dekkers: Awesome. Thank you.
Podcast Information: Bernard Leong (@bernardleong, Linkedin) hosts and produces the show. Proper credits for the intro and end music: "Energetic Sports Drive" and the episode is mixed & edited in both video and audio format by G. Thomas Craig (@gthomascraig, LinkedIn). Here are the links to watch or listen to our podcast.
