MWC Barcelona 2024 | Stephen Rose | IBM

Clips
News
More from MWC Barcelona 2024

Stephen Rose

CEO

Render Networks

IBM’s AI vision: Navigating the future of industry with LLMs

At the forefront of artificial intelligence innovation is the critical balance between strategic implementation and investment returns, as industries worldwide pivot to harness the transformative power of both large and small language models.“There is a massive interest in AI — we’ve come from 2023, where there was a lot of understanding about what’s going on with large language models,” said Stephen Rose (pictured), general manager of global industries at IBM. “I think people trying to get their heads around that

play_circle_outline Exploring the Fascination with AI: An Insightful Conversation with IBM's Stephen Rose about Use Cases and Costs

play_circle_outline Transforming Telco Operators' Mindsets: Widening Access to AI Beyond Specialist Realms

play_circle_outline Use cases in customer care, HR, and code assist

play_circle_outline Co-innovation with customers and GSMA/IBM partnership

play_circle_outline The Pioneering Role of Telco Industry in AI: Transforming Banking and Connecting the World

Info

Stephen Rose | IBM

Stephen Rose

CEO Render Networks

search

(bright music)

>> Welcome back everyone to "theCUBE's" Live coverage here in Barcelona for MWC 2024, Mobile World Congress. I'm John Furrier with "theCUBE", with Dave Vellante and Shelly Kramer. We've got a great guest coming on, General Manager, Global Industry at IBM, Stephen Rose. Stephen, thanks for coming on "theCUBE". First time on "theCUBE", welcome to "theCUBE" alumni.

>> Ah, thank you very much. (John laughs) Absolute pleasure to be here.

>> So first of all, a couple things. General manager, Global Industry. You're seeing all the industries across IBM, which is really a great observation space to look at some of the trends. Here at this show, Telco show also cloud, a lot of edge, a lot of data, really accelerate with the AI tailwind that's coming. So give us a perspective of what's going on for you and your job, set the table, global industries, and then we'll get into this show.

>> Yeah, I mean I think just across all industries, there is a massive interest in AI. Of course. Yeah. So I think we've come from 2023 where there was a lot of understanding about what's going on with large language models. I think people trying to get their heads around that. They're trying to understand what is the efficacy of the use cases, what's the, you know, strategic considerations around the implementation of the technology and particularly what's the cost to do so, what's the return on invested capital for those things? So I think that was really where we've come from. Where we are going to, is suddenly an understanding of, actually LLMs are a big part of the answer, but they're not the only answer. How do we actually look at SLMs, small language models, for example? How do we apply that to different parts of the business and how does that actually then drive different types of use cases, for example? So that's really interesting from a technology lens. And I think actually across different industries there's a still fairly nascent stage, but I think there is a big shift in one thing, which is everybody was originally talking about this as specialist areas. There might be a part of the organization that has a particular, you know, expertise that allows them to actually run new use cases. But now the thinking is, hang on a second, we've got to make this technology available to everybody. Everybody has to go from a user to a content creator. And that's a big shift. 'Cause then you can really apply domain expertise into how to use the technology. So in AI, really exciting stuff and it's really now about applying the technology to domain expertise regardless of the industry.

>> And that's domain expertise, that's data? That specialty data inside the vertical domain is a data opportunity. Hence the excitement.

>> It is. It is actually understanding what data that they actually run with and what data they can get access to. But actually what the domain experts really understand as well is what, what are the things that I struggle with every single day as an employee or in a certain function? Yeah, what are the things I wish I could do that I can't do right now that actually if I'm given the ability to take that type of technology and apply it to that type of data, can I do something different and can I drive a different outcome? So it's a really exciting period.

>> Can you take us inside Stephen, the Telco client mindset. We heard this year, one of the speakers, this morning's keynote, I think it was the Telefonica CEO said, I think it was 55 OTT vendors accounted for 50% of the traffic.

>> Shelly: Yeah. >> Okay. And then last year we heard one of the CEOs, might have been Vodafone, said, they thought that Netflix or OTT vendors should pay a tax, right? And then Netflix countered, well we maybe you should help us pay for the content. And sort of interesting discussions. They also talk about we can't let it happen again with 5G, we have to monetize this time around. All that said, they're really good at connectivity, Telco is. They make great money and relatively, I mean it's high risk, but they've managed that risk, and they know how to manage it. What is the current mindset today and what do you see as their opportunity?

>> Yeah, so I think there's been a massive shift actually. And where the shift has come is in two places. One is Telco operators are really starting to understand, if they want to address the enterprise market, their go-to market model has to change significantly. So actually if you look at the way that they are developing the relationships and their ecosystems, whether they be with application providers or firms like IBM or whether technology firms or whether they're actually with applications providers, they recognize that they need to become more sophisticated in their go to market.

>> Shelly: Yes.

>> I think the second thing is that TCO mindset, that has been plaguing the industry for quite some time now, is starting to become such a burden that they realize, no, we have to change that. So they're looking at actually how do you actually take the different types of use cases, and instead of worrying about cost control and cost management, although critically important, nonetheless it is what can I do with a touch point with a customer that actually uses AI, to be able to convert that into a growth opportunity. And that's the change. And that's a sea change in mindset, because your KPIs start to change, your belief systems in what your core theory of value changes, there's a lot that's with that.

>> It's a whole different business case.

>> Stephen: It is completely different.

>> You know, as opposed to just knocking down the denominator. You're worrying about the numerator.

>> Okay so, on that point, what's the mindset as they go that transition to the new world? Welcome to the party, everybody, it's cloud. Data is critical, they own a lot of data. So where do you see the data opportunities emerging? Because this is the classic case, maybe where that exhaust data exhaust, actually is a net new opportunity. Not just data in the sense of monetizing data, but like the unused data, the traffic data, the operational data. There's a real operational efficiency angle here, as well as a potential business model improvement. Big time. What's your reaction to that?

>> Yeah, I mean, look, again, I think just within operations, there's a sea change in which the operations can actually happen. Imagine that you're able to take, for example, a scrape from what's going on in social media where there's an issue on the network. And you can start to understand based on social media interactions, what people are saying about your network. Can you actually then correlate that with the network data and actually, can you actually pick up triangulation of those two data points that say, actually we know we've got an issue, but more importantly we know who's becoming effective. We know the kind of emotion that that's actually beginning to change. Yeah? So suddenly you start getting insight. And so actually you go from once in a blue moon, NPS type discussion, oh, sorry, NPS scoring, to suddenly you're getting real time NPS, 'cause you understand what the sentiment and what the frustration or what the happiness is, with a particular service. So we've got a complete sea change in the way operations can be done as an example.

>> It's no longer a one snapshot. It's, I've said it a million times. It's a digital representation of your business in real time.

>> Stephen: It is. Yeah yeah, absolutely.

>> You know, I love when we were talking right before the show, you said something that I thought was really cool. You said, you know, here at this show a lot of people are talking about AI. What we're doing is doing AI. So I'd love to hear some just interesting use cases, if you could share anything like that.

>> Yeah, sure. I mean, look, there are three dominant use cases that actually are moving through the industry right now, and across all industries really. One of them is customer care.

>> Yeah.

>> A completely different way in which customer care is happening. And it isn't just about chatbots.

>> No.

>> It's, I mean we typically have been talking about it's chatbots. You get a lot of information through a chat bot, of course it's fantastic. But actually, if you're a customer care agent and you do receive a call, imagine if you're that agent and the AI is actually starting to inference and tell you in real time or very, very near real time, what is the tonality it can observe? What is the tonality in that call? What is the language being used? And then it can actually suggest to you what types of next best action you should be taking as a care agent. Suddenly you're able to think about things and you are not panicking as a customer care agent, because you're under a lot of heat from somebody. But it's also way to able to summarize what may be the historical experience that you had on the screen and maybe a low, a graphical representation of what the experience on the network was. So suddenly you are fully armed as a customer care agent in a way that you weren't before.

>> Well, and that serves, that serves your agents as much as if not, I mean this is a high churn business. Right. Keeping these contact center employees, keeping them happy and Well, I mean, because every day you're dealing with people, customers who are not having their best day. So that AI functionality there makes a big difference in contact center for sure.

>> It does. I mean, imagine, yeah, as I say, I'm the job of a customer care agent. I've done it in my life before.

>> It's hard. >> It's stressful.

>> It's the worst.

>> So you suddenly get all of that taken away from you. The other use cases that we're seeing right now, HR. So massive amounts. In fact IBM we were client zero for what we have is Ask HR, 90% of the ordinary activities that I would go to an HR partner for, which are fairly mundane activities. How do I give somebody a pay rise? How do I make sure that they're in the system properly? How do I onboard an employee? How do I onboard an employee, you know, or off ramp somebody. There's so many things to do, that actually now AI just takes care for me. And not only does it take care of for me in the, you know, sort of normal sense, but also prompts me onto, am I thinking about the second and third things that might come along with a particular decision. And imagine that's not just about Ask HR, that could be Ask Finance, it could be our systems support. It could be any kind of Ask query based use case. So it's fantastic opportunity there.

>> That's wonderful.

>> And what's number three? What's the third big?

>> It's code assist.

>> Oh, of course. >> Code assist is huge, yeah? And again, we've got massive systems in those banking systems, for example, where years and years ago, everything was co coded in COBOL and suddenly we're going to move that to Java. Somebody can get code assisted around that, nobody wants to learn COBOL anymore. But everybody does want the ease of moving into Java.

>> Dave: John loves COBOL. (all laugh) We had this conversation the other day. I said, I hated COBOL. You said you loved COBOL.

>> Oh yeah. It was a one credit lab and I still had to write all that code, but AI could write COBOL for you now. So now you don't need to worry about that. You just, so it's a language.

>> Stephen: Absolutely.

>> You can build a co-pilot to say migrate to the cloud. So I think there's going to be a massive innovation. Again, this is where I'm excited to ask the next question, which is investments in AI. Two part question. One, your investment in AI as you, as a leader, as you bring your value to your customers and help them transform, what are you seeing, that where you're putting your money and your focus and then your customer's investments in the industries? What are they focused on? Is it still, is it creeping up? Is it high, low, medium? Are they doing pilots in production? What's some of the action? Give us some taste of the action there.

>> Yeah, I mean, look, there's a, I mean there's a boatload of action and there's a good reason. I mean, if you look at some of the analyst reports, they're talking about 16 to $18 trillion of new GDP by 2030. So there's a pretty good reason behind it. And if you look at the stock prices of a few of those rather known well known companies, you can see that everybody believes that.

>> Including IBM.

>> Including IBM.

>> But actually what we're seeing a lot of in customers is, is taking the initial use cases, they're building their confidence around those early investments. They will just want to make sure that they get the culture shift in a certain sense. 'Cause there's a culture shift that comes along with it, but also how do we prove out the technology, get everybody's confidence around that. So there's a ton of new use cases early POVs going on. And what we are really focused on is making sure you can do those POVs or POCs within 30 days. And those have got to give you returns straight away as well. So we are really focused on early proof points in the technology. I think the second decision is do you buy, build or borrow your own models? Yeah? And I think there was initial about, oh, we have to have our own model. Then suddenly when you start looking at 180 billion, sorry, 180 billion parameter models and how many, you know, 4,096 GPUs that might take, over a number of different weeks, that'll cost you tens of millions to do that. Suddenly the desire for that is lower. So actually what people are starting to understand is how do I take a large language model and apply that to within use cases to my business and to what extent am I ready to run those types of use cases where maybe the data lineage is not the same confidence I get on a model that I own with my own propriety data. And that might be a small language model at the edge. I'm much more in control of that.

>> It's interesting too. Then the next question is also where do I run those? So I can run as a managed service for say, OpenAI, is one example at the top of the power law. But as you get into specialty models, the small language models as used call them, we call 'em specialty models in our research, they're going to interact with other models. So you see kind of like this API to API concept, models to model integration happening.

>> Stephen: That's right.

>> Where do you run that? You got to run it somewhere. So hybrid, a lot of on-premise activity as process improvement abstractions automate, as you were saying earlier, the mundane heavy lifting, gets moved out of the way. You got to run all this somewhere.

>> Yeah.

>> So you guys are doing a lot of work. Obviously IBM has a cloud strategy. You guys work with Amazon and all the multi clouds. What are customers doing for running and operating this stuff at scale?

>> Yeah, I mean, I think what customers have come to realize is that there's, excuse me, the single approach is hybrid by design. So actually you cannot no longer decide, actually I'm going to run all cloud, all it on prem. You know, there will be different use cases and legit different reasons for that. But the main thing is really that I need to know, have access to all the different clouds I want to have access to. I want to have all different access to data. So hybrid data, and I need hybrid AI because of that. So it's hybrid by design, that's really the way forward. That'll allow you to run the most optimal new models in a way that's TCO and in fact, good for the environment as well, because you're not burning cycles unnecessarily.

>> So Stephen, don't hate me for asking a question that's a little bit of internal plumbing, but I've been watching IBM for many, many, many years and I've always felt like how IBM organizes, because it's such a big complex organization, really matters, matters to your clients. And you've gone through various phases, and Arvind, I think he's talking about at least three pillars, there might be a fourth, but hybrid cloud of consulting and AI. And I for years I felt like the consulting business was sort of the tail wagging the dog. But it seems like there's a much more balanced approach now. You coming in is relatively new, but you've got an industry focus, IBM's industry expertise is phenomenal, best in the industry, along with maybe a couple others. How are you working with that group? Are you in that division? Are you an overlay there? How is it all coming together? And then how do you work with those other key pillars? I saw a couple of red hats just walking by, which is obviously one of those pillars with hybrid cloud and the other parts of the organization, of course AI. How does it all fit together?

>> Yeah, I mean look, we are in a phenomenal opportunity. Oh, sorry, we've got a phenomenal opportunity, and we are in a very lucky position. Excuse me. We are a technology company that's got this phenomenal consulting arm and that's pretty rare actually. But what we also recognize is that our customers want optionality around technology and also consulting partner, if you have a preferred partner. So we can take our technology, we can run it through other GSIs and that it's a perfectly legitimate model. It means that the customers get their favorite SI partner, but they get their favorite technology partner. Of course, if you want to have all of the service, both integration and the technology from IBM, well perfectly great, you know, sorry, we're perfectly happy with that. And that would of course be our preferred model. But really the important thing is that we're a technology company much more, of course around software. We traditionally were hardware orientated. But really we wrap ourselves around four things. Hybrid cloud, automation, security, and AI.

>> Yeah, Dave and I were talking on our CUBE pod, We debate a lot with each other. IBM really took advantage of the AI, because the work you guys have been doing with insights and data and with Watson Y, Watson X, it almost was a perfect storm of opportunity. And that from a timing perspective.

>> Yeah.

>> So how has that changed your role? What and what are you investing in as you look at global industries? Obviously again, we've been saying on "theCUBE" for going back to since 2013, horizontally scalable cloud with vertical specialization in the apps needs to be resolved. That's now happening. So clearly industries, even regulated industries have, well, formed semantic data. So they're actually looking good, like good opportunities too. Cause it's all structured.

>> Yeah.

>> John: And so it's easy to label. (laughs)

>> Yeah.

>> John: So these are like nuances. what's your, what are you working on? How are you attacking this market?

>> Look, I think the most important thing is, everything starts with the use case. So my team really worry about not just the technology, technology comes down the pipe, but actually what we have to figure out is, what is the techno economic benefit for a customer within a particular vertical. So that's really where we focus on for web use case innovation work, yeah? And we, again, there are two elements of that. You either got to be running for cost leadership or you're going to be running for growth. The equation is pretty simple. Yeah? And then what are the return on investments around that? And the second thing is, is you're looking for the characteristics of service that are going to be broadly applicable for more industries. So if you're actually looking at distribution and you're looking at manufacturing and looking at the automation, sorry, automotive sector, there are plenty of similar characteristics. So one use case that you might innovate on will be applicable to multiple industries. So we are looking at those different things.

>> Yeah.

>> That gives us the broadest reach and it also enables multiple industries at once to become more confident around adoption. And then the second element of that is really looking at where does it really turn the needle for both IBM but also for customers of course.

>> So the broader market opportunity, it's a product exercise. You're looking at use cases, you're productizing use cases, identifying ones that sequence across a broader market opportunity, inside a vertical that spans multiple verticals.

>> Stephen: Yeah, absolutely.

>> So can you give an example of one?

>> Yeah, I mean it's, everything starts with why, right? You know, that's a famous, you know, sort of expression, if you don't actually understand why you're implementing the technology or what it's useful for, it's not much use, frankly speaking. So what we are looking at, I mean, again, if we're just looking in telco, for example, we are looking at the, at the cases that I was talking about before. But really we are looking for more advanced cases. How do you connect a system of activities within the value chain? So across the OEMs and the operators, how do you make the observability and the remedial action within a network system actually become more obvious to both parties. Yes. That's a technology provider, like a Nokia or an Ericsson, and the operator. Because the more you can conjoin them, the faster that they can to return to service.

>> You know, I want to get your reaction to something that we've been reporting and researching and it's kind of a new thing, but it's pretty clear in some cases. You brought up productizing use cases. What's interesting is that as foundation models come down the pipe, the intellectual property of your customers is their data and workflows. So they're productizing their values. So what you're actually in the business of doing is productizing workflows. So how should customers think about, first of all, do you believe that statement, that they're having IP in their workflows and data? And how do they create a mode around that and turn that into an opportunity for them? What's your reaction to that?

>> No, so first off, you're absolutely spot on. Look, what's the next platform that we run that is there as an enabler of a customers are, sort of, it's enabler of them to be able to take all of the different use cases that they want to solve for. And the question is, do you want to do it with third party models? Do you want to do it with your own models or do you want to do it with IBM models? Yeah. So first off, we're enabling all of that. And the second thing is, when it comes to IP, well that's a key consideration of whether you build your own model or whether you borrow one. Yeah. So if you're building your own model, then you own the IP and we are the key enabler in that.

>> I think it's a huge opportunity. I think it's going to get bigger and better. I got to ask the final question for me at least, these guys, go one last question each we have, as we run on down of time. What's your schedule look like for the next year? What do you do in your role? Take a minute to explain what you do and put a little plug in for your group.

>> Yeah, no, I mean, look, the biggest thing is co-innovation with customers. Yeah. We've got great customers and we have deep, deep relationships with them. So there's a ton of trust. So we're using that trust to be able to prove out the new use cases. We're taking those, and actually the biggest thing we're doing is this GSMA/IBM partnership, where we're actually allowing all the GSMA partners to take an extended trial on what's next. We're moving those use cases back into the ecosystem. So everybody benefits from it. So a big part of what we want to do is proliferate and actually create a core petition model within the AI community.

>> Shelly: That's awesome.

>> Philanthropy then for the ecosystem.

>> Yeah, absolutely. (John laughs)

>> I won't call-

>> Do you see see Telco as you were talking earlier about how you innovations come out of one industry and then spread to others. Is telco one of the main springs or is it more a recipient? Is there any particular industry that stands out as a mainspring?

>> Actually, it's interesting. I was reading a study the other day, and I wholeheartedly believe this, really, Telco is one of the industries that are going to be leading. They've been greater AI, different forms of AI, for the last 20 years. And it's inherent to the business. So I think, and the other thing is we need to understand, there are two industries, that really make the world spin, banking and connectivity in the Telco industry. When you run those two industries really well, you turn the wheels of all the other industries. So Telco has to be leading, because the demand on them to serve at scale and with the new kind of services that other industries want to provide that rely on Telco, it requires Telco to be leading.

>> And the bets they make, they put the chips in, in this business, and the Telco business is substantial. I mean, you're talking about trillion dollar bets.

>> Stephen: Absolutely.

>> So do you see Telco moving faster now than ever before with AI and the opportunity in front of them or?

>> Yeah, I do actually. And there's some brilliant firms out there, you know, I mean, if I want to put a plug in for Telus, I think they're doing phenomenal stuff. But I also think some of the smaller operators down in, you know, Spark New Zealand, doing some really interesting stuff where they just want to lead on that. So yeah. Really exciting.

>> Spark's coming on later this week.

>> Oh, fantastic. >> Excited to have them.

>> Okay. Yeah.

>> Stephen, thank you so much for spending the time coming on "theCUBE". Really appreciate you. Thanks for sharing all the use case examples and your vision in the industry.

>> Great stuff. >> Stephen: I love being here.

Thank you very much. >> Shelly: Nice job.

>> Okay, Stephen Rose, General Manager, Global Industries for IBM's. Got a great view of the systems. Stephen Rose is here, this is "theCUBE", with live coverage after this short break, we'll be right back. (bright music)

Stephen Rose | IBM

search

(bright music)

>> Welcome back everyone to "theCUBE's" Live coverage here in Barcelona for MWC 2024, Mobile World Congress. I'm John Furrier with "theCUBE", with Dave Vellante and Shelly Kramer. We've got a great guest coming on, General Manager, Global Industry at IBM, Stephen Rose. Stephen, thanks for coming on "theCUBE". First time on "theCUBE", welcome to "theCUBE" alumni.

>> Ah, thank you very much. (John laughs) Absolute pleasure to be here.

>> So first of all, a couple things. General manager, Global Industry. You're seeing all the industries across IBM, which is really a great observation space to look at some of the trends. Here at this show, Telco show also cloud, a lot of edge, a lot of data, really accelerate with the AI tailwind that's coming. So give us a perspective of what's going on for you and your job, set the table, global industries, and then we'll get into this show.

>> Yeah, I mean I think just across all industries, there is a massive interest in AI. Of course. Yeah. So I think we've come from 2023 where there was a lot of understanding about what's going on with large language models. I think people trying to get their heads around that. They're trying to understand what is the efficacy of the use cases, what's the, you know, strategic considerations around the implementation of the technology and particularly what's the cost to do so, what's the return on invested capital for those things? So I think that was really where we've come from. Where we are going to, is suddenly an understanding of, actually LLMs are a big part of the answer, but they're not the only answer. How do we actually look at SLMs, small language models, for example? How do we apply that to different parts of the business and how does that actually then drive different types of use cases, for example? So that's really interesting from a technology lens. And I think actually across different industries there's a still fairly nascent stage, but I think there is a big shift in one thing, which is everybody was originally talking about this as specialist areas. There might be a part of the organization that has a particular, you know, expertise that allows them to actually run new use cases. But now the thinking is, hang on a second, we've got to make this technology available to everybody. Everybody has to go from a user to a content creator. And that's a big shift. 'Cause then you can really apply domain expertise into how to use the technology. So in AI, really exciting stuff and it's really now about applying the technology to domain expertise regardless of the industry.

>> And that's domain expertise, that's data? That specialty data inside the vertical domain is a data opportunity. Hence the excitement.

>> It is. It is actually understanding what data that they actually run with and what data they can get access to. But actually what the domain experts really understand as well is what, what are the things that I struggle with every single day as an employee or in a certain function? Yeah, what are the things I wish I could do that I can't do right now that actually if I'm given the ability to take that type of technology and apply it to that type of data, can I do something different and can I drive a different outcome? So it's a really exciting period.

>> Can you take us inside Stephen, the Telco client mindset. We heard this year, one of the speakers, this morning's keynote, I think it was the Telefonica CEO said, I think it was 55 OTT vendors accounted for 50% of the traffic.

>> Shelly: Yeah. >> Okay. And then last year we heard one of the CEOs, might have been Vodafone, said, they thought that Netflix or OTT vendors should pay a tax, right? And then Netflix countered, well we maybe you should help us pay for the content. And sort of interesting discussions. They also talk about we can't let it happen again with 5G, we have to monetize this time around. All that said, they're really good at connectivity, Telco is. They make great money and relatively, I mean it's high risk, but they've managed that risk, and they know how to manage it. What is the current mindset today and what do you see as their opportunity?

>> Yeah, so I think there's been a massive shift actually. And where the shift has come is in two places. One is Telco operators are really starting to understand, if they want to address the enterprise market, their go-to market model has to change significantly. So actually if you look at the way that they are developing the relationships and their ecosystems, whether they be with application providers or firms like IBM or whether technology firms or whether they're actually with applications providers, they recognize that they need to become more sophisticated in their go to market.

>> Shelly: Yes.

>> I think the second thing is that TCO mindset, that has been plaguing the industry for quite some time now, is starting to become such a burden that they realize, no, we have to change that. So they're looking at actually how do you actually take the different types of use cases, and instead of worrying about cost control and cost management, although critically important, nonetheless it is what can I do with a touch point with a customer that actually uses AI, to be able to convert that into a growth opportunity. And that's the change. And that's a sea change in mindset, because your KPIs start to change, your belief systems in what your core theory of value changes, there's a lot that's with that.

>> It's a whole different business case.

>> Stephen: It is completely different.

>> You know, as opposed to just knocking down the denominator. You're worrying about the numerator.

>> Okay so, on that point, what's the mindset as they go that transition to the new world? Welcome to the party, everybody, it's cloud. Data is critical, they own a lot of data. So where do you see the data opportunities emerging? Because this is the classic case, maybe where that exhaust data exhaust, actually is a net new opportunity. Not just data in the sense of monetizing data, but like the unused data, the traffic data, the operational data. There's a real operational efficiency angle here, as well as a potential business model improvement. Big time. What's your reaction to that?

>> Yeah, I mean, look, again, I think just within operations, there's a sea change in which the operations can actually happen. Imagine that you're able to take, for example, a scrape from what's going on in social media where there's an issue on the network. And you can start to understand based on social media interactions, what people are saying about your network. Can you actually then correlate that with the network data and actually, can you actually pick up triangulation of those two data points that say, actually we know we've got an issue, but more importantly we know who's becoming effective. We know the kind of emotion that that's actually beginning to change. Yeah? So suddenly you start getting insight. And so actually you go from once in a blue moon, NPS type discussion, oh, sorry, NPS scoring, to suddenly you're getting real time NPS, 'cause you understand what the sentiment and what the frustration or what the happiness is, with a particular service. So we've got a complete sea change in the way operations can be done as an example.

>> It's no longer a one snapshot. It's, I've said it a million times. It's a digital representation of your business in real time.

>> Stephen: It is. Yeah yeah, absolutely.

>> You know, I love when we were talking right before the show, you said something that I thought was really cool. You said, you know, here at this show a lot of people are talking about AI. What we're doing is doing AI. So I'd love to hear some just interesting use cases, if you could share anything like that.

>> Yeah, sure. I mean, look, there are three dominant use cases that actually are moving through the industry right now, and across all industries really. One of them is customer care.

>> Yeah.

>> A completely different way in which customer care is happening. And it isn't just about chatbots.

>> No.

>> It's, I mean we typically have been talking about it's chatbots. You get a lot of information through a chat bot, of course it's fantastic. But actually, if you're a customer care agent and you do receive a call, imagine if you're that agent and the AI is actually starting to inference and tell you in real time or very, very near real time, what is the tonality it can observe? What is the tonality in that call? What is the language being used? And then it can actually suggest to you what types of next best action you should be taking as a care agent. Suddenly you're able to think about things and you are not panicking as a customer care agent, because you're under a lot of heat from somebody. But it's also way to able to summarize what may be the historical experience that you had on the screen and maybe a low, a graphical representation of what the experience on the network was. So suddenly you are fully armed as a customer care agent in a way that you weren't before.

>> Well, and that serves, that serves your agents as much as if not, I mean this is a high churn business. Right. Keeping these contact center employees, keeping them happy and Well, I mean, because every day you're dealing with people, customers who are not having their best day. So that AI functionality there makes a big difference in contact center for sure.

>> It does. I mean, imagine, yeah, as I say, I'm the job of a customer care agent. I've done it in my life before.

>> It's hard. >> It's stressful.

>> It's the worst.

>> So you suddenly get all of that taken away from you. The other use cases that we're seeing right now, HR. So massive amounts. In fact IBM we were client zero for what we have is Ask HR, 90% of the ordinary activities that I would go to an HR partner for, which are fairly mundane activities. How do I give somebody a pay rise? How do I make sure that they're in the system properly? How do I onboard an employee? How do I onboard an employee, you know, or off ramp somebody. There's so many things to do, that actually now AI just takes care for me. And not only does it take care of for me in the, you know, sort of normal sense, but also prompts me onto, am I thinking about the second and third things that might come along with a particular decision. And imagine that's not just about Ask HR, that could be Ask Finance, it could be our systems support. It could be any kind of Ask query based use case. So it's fantastic opportunity there.

>> That's wonderful.

>> And what's number three? What's the third big?

>> It's code assist.

>> Oh, of course. >> Code assist is huge, yeah? And again, we've got massive systems in those banking systems, for example, where years and years ago, everything was co coded in COBOL and suddenly we're going to move that to Java. Somebody can get code assisted around that, nobody wants to learn COBOL anymore. But everybody does want the ease of moving into Java.

>> Dave: John loves COBOL. (all laugh) We had this conversation the other day. I said, I hated COBOL. You said you loved COBOL.

>> Oh yeah. It was a one credit lab and I still had to write all that code, but AI could write COBOL for you now. So now you don't need to worry about that. You just, so it's a language.

>> Stephen: Absolutely.

>> You can build a co-pilot to say migrate to the cloud. So I think there's going to be a massive innovation. Again, this is where I'm excited to ask the next question, which is investments in AI. Two part question. One, your investment in AI as you, as a leader, as you bring your value to your customers and help them transform, what are you seeing, that where you're putting your money and your focus and then your customer's investments in the industries? What are they focused on? Is it still, is it creeping up? Is it high, low, medium? Are they doing pilots in production? What's some of the action? Give us some taste of the action there.

>> Yeah, I mean, look, there's a, I mean there's a boatload of action and there's a good reason. I mean, if you look at some of the analyst reports, they're talking about 16 to $18 trillion of new GDP by 2030. So there's a pretty good reason behind it. And if you look at the stock prices of a few of those rather known well known companies, you can see that everybody believes that.

>> Including IBM.

>> Including IBM.

>> But actually what we're seeing a lot of in customers is, is taking the initial use cases, they're building their confidence around those early investments. They will just want to make sure that they get the culture shift in a certain sense. 'Cause there's a culture shift that comes along with it, but also how do we prove out the technology, get everybody's confidence around that. So there's a ton of new use cases early POVs going on. And what we are really focused on is making sure you can do those POVs or POCs within 30 days. And those have got to give you returns straight away as well. So we are really focused on early proof points in the technology. I think the second decision is do you buy, build or borrow your own models? Yeah? And I think there was initial about, oh, we have to have our own model. Then suddenly when you start looking at 180 billion, sorry, 180 billion parameter models and how many, you know, 4,096 GPUs that might take, over a number of different weeks, that'll cost you tens of millions to do that. Suddenly the desire for that is lower. So actually what people are starting to understand is how do I take a large language model and apply that to within use cases to my business and to what extent am I ready to run those types of use cases where maybe the data lineage is not the same confidence I get on a model that I own with my own propriety data. And that might be a small language model at the edge. I'm much more in control of that.

>> It's interesting too. Then the next question is also where do I run those? So I can run as a managed service for say, OpenAI, is one example at the top of the power law. But as you get into specialty models, the small language models as used call them, we call 'em specialty models in our research, they're going to interact with other models. So you see kind of like this API to API concept, models to model integration happening.

>> Stephen: That's right.

>> Where do you run that? You got to run it somewhere. So hybrid, a lot of on-premise activity as process improvement abstractions automate, as you were saying earlier, the mundane heavy lifting, gets moved out of the way. You got to run all this somewhere.

>> Yeah.

>> So you guys are doing a lot of work. Obviously IBM has a cloud strategy. You guys work with Amazon and all the multi clouds. What are customers doing for running and operating this stuff at scale?

>> Yeah, I mean, I think what customers have come to realize is that there's, excuse me, the single approach is hybrid by design. So actually you cannot no longer decide, actually I'm going to run all cloud, all it on prem. You know, there will be different use cases and legit different reasons for that. But the main thing is really that I need to know, have access to all the different clouds I want to have access to. I want to have all different access to data. So hybrid data, and I need hybrid AI because of that. So it's hybrid by design, that's really the way forward. That'll allow you to run the most optimal new models in a way that's TCO and in fact, good for the environment as well, because you're not burning cycles unnecessarily.

>> So Stephen, don't hate me for asking a question that's a little bit of internal plumbing, but I've been watching IBM for many, many, many years and I've always felt like how IBM organizes, because it's such a big complex organization, really matters, matters to your clients. And you've gone through various phases, and Arvind, I think he's talking about at least three pillars, there might be a fourth, but hybrid cloud of consulting and AI. And I for years I felt like the consulting business was sort of the tail wagging the dog. But it seems like there's a much more balanced approach now. You coming in is relatively new, but you've got an industry focus, IBM's industry expertise is phenomenal, best in the industry, along with maybe a couple others. How are you working with that group? Are you in that division? Are you an overlay there? How is it all coming together? And then how do you work with those other key pillars? I saw a couple of red hats just walking by, which is obviously one of those pillars with hybrid cloud and the other parts of the organization, of course AI. How does it all fit together?

>> Yeah, I mean look, we are in a phenomenal opportunity. Oh, sorry, we've got a phenomenal opportunity, and we are in a very lucky position. Excuse me. We are a technology company that's got this phenomenal consulting arm and that's pretty rare actually. But what we also recognize is that our customers want optionality around technology and also consulting partner, if you have a preferred partner. So we can take our technology, we can run it through other GSIs and that it's a perfectly legitimate model. It means that the customers get their favorite SI partner, but they get their favorite technology partner. Of course, if you want to have all of the service, both integration and the technology from IBM, well perfectly great, you know, sorry, we're perfectly happy with that. And that would of course be our preferred model. But really the important thing is that we're a technology company much more, of course around software. We traditionally were hardware orientated. But really we wrap ourselves around four things. Hybrid cloud, automation, security, and AI.

>> Yeah, Dave and I were talking on our CUBE pod, We debate a lot with each other. IBM really took advantage of the AI, because the work you guys have been doing with insights and data and with Watson Y, Watson X, it almost was a perfect storm of opportunity. And that from a timing perspective.

>> Yeah.

>> So how has that changed your role? What and what are you investing in as you look at global industries? Obviously again, we've been saying on "theCUBE" for going back to since 2013, horizontally scalable cloud with vertical specialization in the apps needs to be resolved. That's now happening. So clearly industries, even regulated industries have, well, formed semantic data. So they're actually looking good, like good opportunities too. Cause it's all structured.

>> Yeah.

>> John: And so it's easy to label. (laughs)

>> Yeah.

>> John: So these are like nuances. what's your, what are you working on? How are you attacking this market?

>> Look, I think the most important thing is, everything starts with the use case. So my team really worry about not just the technology, technology comes down the pipe, but actually what we have to figure out is, what is the techno economic benefit for a customer within a particular vertical. So that's really where we focus on for web use case innovation work, yeah? And we, again, there are two elements of that. You either got to be running for cost leadership or you're going to be running for growth. The equation is pretty simple. Yeah? And then what are the return on investments around that? And the second thing is, is you're looking for the characteristics of service that are going to be broadly applicable for more industries. So if you're actually looking at distribution and you're looking at manufacturing and looking at the automation, sorry, automotive sector, there are plenty of similar characteristics. So one use case that you might innovate on will be applicable to multiple industries. So we are looking at those different things.

>> Yeah.

>> That gives us the broadest reach and it also enables multiple industries at once to become more confident around adoption. And then the second element of that is really looking at where does it really turn the needle for both IBM but also for customers of course.

>> So the broader market opportunity, it's a product exercise. You're looking at use cases, you're productizing use cases, identifying ones that sequence across a broader market opportunity, inside a vertical that spans multiple verticals.

>> Stephen: Yeah, absolutely.

>> So can you give an example of one?

>> Yeah, I mean it's, everything starts with why, right? You know, that's a famous, you know, sort of expression, if you don't actually understand why you're implementing the technology or what it's useful for, it's not much use, frankly speaking. So what we are looking at, I mean, again, if we're just looking in telco, for example, we are looking at the, at the cases that I was talking about before. But really we are looking for more advanced cases. How do you connect a system of activities within the value chain? So across the OEMs and the operators, how do you make the observability and the remedial action within a network system actually become more obvious to both parties. Yes. That's a technology provider, like a Nokia or an Ericsson, and the operator. Because the more you can conjoin them, the faster that they can to return to service.

>> You know, I want to get your reaction to something that we've been reporting and researching and it's kind of a new thing, but it's pretty clear in some cases. You brought up productizing use cases. What's interesting is that as foundation models come down the pipe, the intellectual property of your customers is their data and workflows. So they're productizing their values. So what you're actually in the business of doing is productizing workflows. So how should customers think about, first of all, do you believe that statement, that they're having IP in their workflows and data? And how do they create a mode around that and turn that into an opportunity for them? What's your reaction to that?

>> No, so first off, you're absolutely spot on. Look, what's the next platform that we run that is there as an enabler of a customers are, sort of, it's enabler of them to be able to take all of the different use cases that they want to solve for. And the question is, do you want to do it with third party models? Do you want to do it with your own models or do you want to do it with IBM models? Yeah. So first off, we're enabling all of that. And the second thing is, when it comes to IP, well that's a key consideration of whether you build your own model or whether you borrow one. Yeah. So if you're building your own model, then you own the IP and we are the key enabler in that.

>> I think it's a huge opportunity. I think it's going to get bigger and better. I got to ask the final question for me at least, these guys, go one last question each we have, as we run on down of time. What's your schedule look like for the next year? What do you do in your role? Take a minute to explain what you do and put a little plug in for your group.

>> Yeah, no, I mean, look, the biggest thing is co-innovation with customers. Yeah. We've got great customers and we have deep, deep relationships with them. So there's a ton of trust. So we're using that trust to be able to prove out the new use cases. We're taking those, and actually the biggest thing we're doing is this GSMA/IBM partnership, where we're actually allowing all the GSMA partners to take an extended trial on what's next. We're moving those use cases back into the ecosystem. So everybody benefits from it. So a big part of what we want to do is proliferate and actually create a core petition model within the AI community.

>> Shelly: That's awesome.

>> Philanthropy then for the ecosystem.

>> Yeah, absolutely. (John laughs)

>> I won't call-

>> Do you see see Telco as you were talking earlier about how you innovations come out of one industry and then spread to others. Is telco one of the main springs or is it more a recipient? Is there any particular industry that stands out as a mainspring?

>> Actually, it's interesting. I was reading a study the other day, and I wholeheartedly believe this, really, Telco is one of the industries that are going to be leading. They've been greater AI, different forms of AI, for the last 20 years. And it's inherent to the business. So I think, and the other thing is we need to understand, there are two industries, that really make the world spin, banking and connectivity in the Telco industry. When you run those two industries really well, you turn the wheels of all the other industries. So Telco has to be leading, because the demand on them to serve at scale and with the new kind of services that other industries want to provide that rely on Telco, it requires Telco to be leading.

>> And the bets they make, they put the chips in, in this business, and the Telco business is substantial. I mean, you're talking about trillion dollar bets.

>> Stephen: Absolutely.

>> So do you see Telco moving faster now than ever before with AI and the opportunity in front of them or?

>> Yeah, I do actually. And there's some brilliant firms out there, you know, I mean, if I want to put a plug in for Telus, I think they're doing phenomenal stuff. But I also think some of the smaller operators down in, you know, Spark New Zealand, doing some really interesting stuff where they just want to lead on that. So yeah. Really exciting.

>> Spark's coming on later this week.

>> Oh, fantastic. >> Excited to have them.

>> Okay. Yeah.

>> Stephen, thank you so much for spending the time coming on "theCUBE". Really appreciate you. Thanks for sharing all the use case examples and your vision in the industry.

>> Great stuff. >> Stephen: I love being here.

Thank you very much. >> Shelly: Nice job.

>> Okay, Stephen Rose, General Manager, Global Industries for IBM's. Got a great view of the systems. Stephen Rose is here, this is "theCUBE", with live coverage after this short break, we'll be right back. (bright music)