Reinventing IT for the AI Era in Telecom

Hi. This is Mitch Wagner, chief analyst AI data center and cloud for Fierce Network Research. I am here today for a conversation with TM Forum's Willie Stegmann. He is EVP composable IT and ecosystems for TM Forum. Welcome, Willie.

But I think this one is different. I think you would agree with me that the scale, the magnitude, the velocity of what's happened the last two years around AI is just absolutely phenomenal. I haven't seen that in the last 35 years of my career, certainly not what we've experienced.

And if you look at the pressure that the CIO is under, both externally as well as internally, just reflect on that for a minute. If you think about the external pressure, most of the markets, the telco industry is very mature. Growth rates are really hard to come by. It's heavily regulated. It's very, very competitive. And so there's a lot of pressure on the IT organization to enable the organization to remain relevant under those circumstances.

We have lots of risk from a cyber point of view that's still there. Now we have the additional responsibility that are making AI responsible and safe for internal as well as customer consumption. And then the internal pressures are also significant. IT is still seen as slow, cumbersome, and a lot of legacy that's keeping the organization back in terms of being able to really be fast, responsive, and competitive.

There's increasingly what we are hearing, Mitch, targets being put on the CIOs to realize savings from AI, not to only enable AI for the organization, but also to use AI in the IT organization to run things more efficiently, more effectively, and to put savings on the table to be able to afford all of the AI investments that's making. Not to mention things like the skills challenge that we've got, the importance of engineering in the broader sense.

So if you look at all of these forces that's being imposed on a CIO, we believe that it's necessary to really call for a rethink, a reinvention of the IT organization from the foundations all the way up to how we realize value. And that's our call to action. And that's what we're going to take to DTW this year in Copenhagen in June. And we believe there's urgency for CSP CIOs and operators to act.

That's interesting. So that does lead us to our next question, which is that AI is moving from experimentation to core operations. But as that happens, many CSPs are finding that their existing architectures just can't scale safely or consistently. What does an AI native foundation actually require in practice and why does common architecture matter?

And as TM Forum, we've been very vocal over the last close to 10 years, if not longer, that we believe it's really critical for the industry to adopt a true north architecture of reference, which is our open digital architecture and associated open APIs. I think it's been, Mitch, very, very critical and very supportive. I mean, when I worked at an operator as well, we used ODA very effectively, I think, to guidepost and to signpost our digital transformation journey.

The opportunity that we've got now is, and also the reality, if you have not succeeded in terms of making your architectures more open, more composable, more agile, you're now layering AI on top of this very complex legacy, messy, slow, cumbersome architecture, it's very difficult for you to give the business the outcomes that they desire, lower cost, velocity, responsiveness, enablement.

So what we believe is still is that the call to action is for CIOs to really think about their architectures, use ODA as your architecture of reference. But the important development here, Mitch, is that with the support from the members, because that's where TM Forum really comes into play, we don't do it ourselves, we do it for the members, with the members, is to evolve the open digital architecture to become intelligent and to become AI native. And that is what we're currently busy with.

If you look at what we're going to demonstrate at DTW is, we're going to not talk about it in slideware, physically we'll demonstrate the code associated with how do you evolve your architecture to become more open, more faster velocity, and also more intelligent.

So I think the good news, Mitch, is, is we don't believe you have to put a cross through everything and start from scratch. If you've walked the journey with the TM Forum, if you introduced open digital architecture components, open APIs, we believe you'll continue with that journey as we evolve the architecture to become more intelligent and really embrace the AI era that we spoke about earlier.

How do we close the gap between ambition and reality? Many operators have clear AI or autonomy strategy, but not a lot of them are seeing consistent results at scale. Where do CSPs typically get stuck when moving from design to production and what actually helps them break through that execution barrier?

It's such a good question, Mitch. And I think the answer is quite nuanced. A lot of research at the moment. If you open, if you go LinkedIn, you'll see all the big research houses are grappling with this question.

One of the things that we've seen come out as a consistent theme is going from the first pilot, the first few test examples, use cases to really to scale, that's proving to be very, very difficult. And because the technologies works, I think yes, there's the discussions around privacy and accuracy and hallucination, et cetera. Let's just park that for the moment. We know the technology works, it can scale. There's no doubt. Yes, it costs a lot of money as you scale and you use more tokens, it costs money, but I think that the challenge to scale is a lot more fundamental. I think it's around culture, it's around people, it's around process, it's around risk appetite, it's around prioritization. Is this something that you want to do? Does it compete against other initiatives? So going from something small to something that's enterprise wide on paper looks easy, in practice, what we've seen has been really, really difficult.

There's also, Mitch, I think a tendency and a danger of lock-in, technology lock-in. So because everybody wants to move quickly, they want to scale, you're then partnering with one or two technology companies to allow that scaling to happen. And in the process, you actually lock yourself in. And so I think there's really, really important dimension here of openness when you scale to make sure that you don't get unnecessarily lock-in, both from a technology commercial point of view as well.

So I think you need proper governance. The examples that we've seen where scaling has happened successfully, typically the executive teams all the way to CEO level have been very involved in terms of selecting the scalable use cases, understanding the business benefits. The CIO, CTO have been heavily involved in terms of the technology, how do you actually do that? And if it's owned at that strategic level, the chances of success of the scaling happening is so much higher. And the reality is this environment, it's better to do fewer things better than a lot of things scale at the same time. We've seen that being really problematic.

Last point, I think most of the early success has been in the customer engagement channels, typically your contact centers where we know we have had machine learning for a number of years to augment that with LLM technology is possible. And so we've seen, I think, quick uptake there and also associated commercial, customer, and internal benefits. So I think that's been most successful from what we've seen so far, but clearly the scaling is going much, much broader and deeper beyond that at the moment.

That is an interesting insight and that does lead us to our final question, which is what does success look like for CIOs? In two or three years when CIOs look back, what tangible outcomes will tell them they've succeeded in reinventing IT beyond automation for automation's sake? What should they be measuring?

Super question, Mitch. So we are currently working through some of the numbers that I sense I would say three critical dimensions.

I think that the first one is, I would say the first one is internal. The second, third one is more external. Let's start with the internal one. The CIO needs to prove after three years that they've made significant progress to remove legacy. We believe that there's a challenge to actually get to zero legacy. Now, that may seem naive and impossible, but I believe with leveraging AI, it's definitely possible to get to a point where your dependency on messy, hard, ugly legacy really falls away. So I think that's really critical.

Operational excellence in the context of ongoing stability and very importantly, increasing velocity, time to market, I think for IT will be a critical measure. And what we want to do in the forum is to signpost and give a sense of what constitutes significant progress from today versus what is there in three years from now.

And then cost savings. I mentioned earlier, the CIOs being under pressure from the board, from the executive team, from the CEO to say, "We need you to help us to pay for all the AI that we are deploying, and you're going to have to do that out of leveraging AI in your own environment and do things smarter and better." And I think that's easy to measure, right? Your IT cost as a percentage of your service revenue could be a potential metric and how that shifts by 20, 30% in the next three years I think is indicative.

The other two then is externally. I would say the first one is customer experience. We believe that there's both a threat and opportunity here for CSPs. The opportunity is that you take all the good work that we've done through the digital transformation to serve customers in an omnichannel way across all the different channels, including things like app and web, and you make those experiences really intelligent, really agentic.

We firmly believe that it's critical that operators accept that if they don't do it, their competitors will and you will lose customers. It's very possible that in future customers will interact with your company as an organization through agents as opposed to through humans. And if you can do that in an efficient and effective way after three years, I think a very important subtext to your reinvention journey.

And then the third one is the extent to which as the CIO, as the IT organization, you've been able to enable exploiting what we call ecosystem opportunities or adjacent opportunities outside of the core mobile offering. And there are a few that's on our radar at the moment, Mitch. The first one is like satellite ODA, leveraging ODA in the context of satellite, we've seen a lot of announcements where satellite and terrestrial operators are working together to seamlessly give customers a much better coverage irrespective of geography, in the mountains or in the lakes or far removed from the cities where you don't have towers at the moment through the terrestrial service.

Secondly is fintech. A lot of our markets, fintech is a very critical revenue enabler, very substantial revenue enabler. IT must be seen to be advancing that as well. Then you have things like wholesale broadband, ongoing monetization of APIs and MVNOs, don't forget about MVNOs, that's still a very, very important topic.

So in short, after three years, if you've shown real progress to move legacy, bring the costs down, serve customers in a very agentic and intelligent way and then lastly, enable the business to really attack the adjacent opportunities. I would say you've succeeded as a CIO to have reinvented your IT organization.

And that concludes this conversation. Thank you very much for joining us and sharing your insights, Willie.