International Business Machines Corporation (IBM) Earnings Call Transcript & Summary

Good afternoon, everyone, and welcome. I'm Toni Sacconaghi, Bernstein's IT hardware analyst. And I'm very pleased today to welcome Arvind Krishna, Chairman and CEO of IBM. Arvind has been at IBM for 30 years and was previously the SVP of Cloud & Cognitive Software. He also headed IBM Research and the Systems and Technology group. He was appointed CEO of IBM in April 2020, so just over 14 months ago, and we're thrilled to have Arvind join us this afternoon. Welcome, Arvind, and thanks for participating in Bernstein's SDC.

Toni, it's a pleasure to be here speaking with you and to your audience.

So as I mentioned, you've been CEO for 14 months. What has been perhaps a pleasant surprise, and what maybe has been tougher than you thought reflecting on this initial period?

So Toni, I'll begin maybe with the second part of your question. So what's tougher is what I think was the unexpected part. So I was named on January 30, 2020. And I think the pandemic was not even a glimmer in our eyes at that point. Hindsight says maybe we were all kind of blind and deaf to it because I think the month of January in China had already got that. It's not so much the health impact of the pandemic, but the economic impact of the pandemic has been a surprise to deal with, a negative surprise. I wouldn't call it a positive surprise and so having to worry a lot more about what could happen to the economy, having to worry about having everyone work effectively but remote. We are really ecstatic that our people have had 98% productivity of what they might have had otherwise during this time. But that was all definitely a negative surprise. And 2020's GDP impact around the globe, we had certainly felt them. We are not immune to them. So that was the biggest negative. On the other side, you said the positive. I think the positive for me has been the speed at which we've been able to get on our strategy, our strategy being our hybrid cloud and AI. So you heard us talk late last year about spinning out Kyndryl. We now have a name. So we're spinning that out. Managed to move with speed on getting 12 acquisitions under our belt. That was part of my commitment that we're going to spend on both organic and inorganic. And we have begun a number of go-to-market actions that are not quite completed, but that are underway. So I think the willingness of our clients and our employees to embrace a much more focused strategy was the positive surprise. And I think the pandemic and its economic impacts was the big negative.

Great. Well, maybe we can talk a little bit more about the strategy and vision for IBM. So you've been very consistent, very deliberate in saying that IBM is a Cloud & Cognitive or cloud and AI company going forward. Maybe you can start with, for investors who know you well and some perhaps who don't know you as well, to articulate really what is AI at IBM, how big is it within IBM and where might investors see it either on the income statement or in IBM's businesses.

Yes. So Toni, I'll address the AI question, and then we'll come back to hybrid cloud at some point, I hope, during the conversation, [ ourselves ]. So I'm going to first mention 2 things that really -- because you've got to begin with a business focus and a financial lens on where is AI. So AI, I think it was PwC who did this estimate, by 2030 will cost $16 trillion or unlock $16 trillion in global productivity. And then you say, well, why? Because we're also going to be generating 175 zettabytes of data each year. When you generate that much data, the only technology we know that can unlock value and get insights from that much data is artificial intelligence, okay? So a huge price to be had by all our clients on the $16 trillion, our clients and governments, convenience felt by the end user, and given the size of data, there is no alternate technology. So now you begin to put that together and you say, so where do you see this. So I want to be clear to everybody. Watson is alive and well. But where are you going to see it? We had our annual user conference last month, Think. And I'll give you a great example and then turn from their to Toni's question of where all do you see it. So Karen Lynch, the CEO of CVS Health, was there and she spoke. And she described how, in December, CVS began to launch a vaccine management, in their case, mostly North America and the United States. And they came to us and said, look, we are going to do tens of millions of vaccines every quarter. We're going to get 20 million, 30 million, 40 million calls. Do I need to staff up a call center that can handle all those calls, answer people's queries? Or is there technology that can go do that? So we stood up a Watson conversational agent that now has to do what, what is CVS' protocol, what is each of the 50 states in the U.S. because they all had their own regulations, what are people's conditions, and how do you answer the question of can I come in for a vaccine appointment, am I eligible, is it going to get paid for, when can I get an appointment. Over 70% of all those calls were handled by a Watson agent. Now if you think about that, 17 million calls, I think that is per quarter. You now extrapolate that over the year. That is a great success point. Now you can begin to get into doing things like that for different things other than vaccines. Then you can get into, can I do it for hiring, can I do it for people movement inside, think about HR. So the number of enterprise processes where you can begin to apply these technologies, I think, is incredible. That's how you unlock the $16 trillion in global productivity I spoke of because you can do your mental math and say, okay, you just talked about a few tens of millions for one use case at one customer. So I kind of go, but there are 10 such use cases at this customer, and then you can go across thousands of clients with those use cases. Now where are you going to see it show up? You're going to see it show up, for example, in how we do predictive maintenance inside products like Maximo on how can you predict when some parts might need maintenance ahead of a time schedule or a person doing a look-see. You can begin to look at other enterprise use cases around can I talk to an agent as opposed to coming on my sales force screener. This is something we are doing internally. I can have another agent that picks up between my mail, my internal pricing tools and sales force to help make a client-facing person a lot more productive. That is the places where we are going to do AI and bringing AI to the enterprise. So Toni, as you think about that, certainly, a lot of data informs how you build your base model. But then what we do at CVS with their data, their question and answers, their protocols, we keep to them. And I call that small data because that's not millions of questions and millions of Q&A that you're doing. You're learning from every interaction, of course, to improve it, but that model stays with CVS. We never take it anywhere else. So this ability to begin with some base technologies, but then look a client in the eye and say, your data is only yours, any insight from that is only yours, I think is somewhat unique to us. And the ability to do that in a regulated world, in a place where people care about privacy incredibly and we can look them in the eye and guarantee that we are never going to take that data somewhere else is, I think, a real advantage. And doing that across enterprises is our focus. We are not going to be investing in consumer AI. That's not, I think, what we are best at. Many others are going to be great at doing consumer AI.

Right. So -- but Arvind, in that CVS example, and I was vaccinated at CVS, so I did have some of that AI experience and it's pretty seamless. But where is IBM recognizing revenue? Is it largely in the GBS business? Are you recognizing it also in your Cloud & Cognitive Software business? And how do we just think about the revenue recognition model for AI because I -- and has it changed? Because initially, it felt like Watson was really about core technologies. And then IBM acquired a number of firms in both health care and Promontory Financial and look like it was going to market in a very more siloed vertical market approach for AI. We've heard less about Watson Financial and Watson Health, and how you describe the CVS example felt almost like you were going back to the roots, which is we have core technologies, we'll deploy them. So 2 questions, where would -- where is revenue likely to be most recognized in terms of Watson or AI solutions going forward? And is your view of how Watson will be brought to market consistent with developing vertical market capabilities? Or is it less so?

Great questions, Toni, and thank you because it allows me to explain. And I will call it, it is a change in the strategy, so let me acknowledge that, from a few years ago. So you're right. A few years ago, we tried to bring these technologies to market through verticals. Health, financial being 2 examples, [ and I can go ] on marketing, perhaps, being 1/3. We are not going to be doing the verticals. I think that a much better market for IBM as actually the last year is proving out is to bring core technologies and let the verticals be done by the client, in conjunction with them. So to first unpack your question. You're going to see the AI revenue show up in 2 places, absolutely in GBS, who does thousands of projects with clients on these. Actually, the CVS example was a GBS implementation, by the way, I'll like claim in 3 weeks, so it shows that the technologies are pretty mature and pretty ready. To go from day 1 to the call -- to the Watson answering calls in 3 weeks, I think is pretty remarkable, actually. So that's [ a great thing ]. But then there is a longer-term revenue stream that will show up in Cloud & Cognitive Software, specifically within the cloud and data platform segment, to get probably even more specific than your audience might want, but I know, Toni, you follow that. And that will show up mostly in a as a Service. So in the example I talked about, actually, the clients don't pay anything upfront. They pay according to the number of calls, API calls, if we want to use their jargon that they make. So in that case, it was, whatever, I think, 17 million API calls in the first quarter, and that will then show up as revenue in that segment. And GBS, of course, as they do the work, to do the upfront training, the install, sometimes the run, not always the run, are the 2 primary places you'll see the revenue show up. So in this case, whatever you consume off the cloud, et cetera, will be inside the software number. Now we do allow Watson to run on-premise as well. And then the client gets the ability to just pay according to the number of API calls, and they are using their own infrastructure in that case. So that is also a difference there. So I think I answered where the revenue shows up, and it's in both places. We believe it's a really big driver of intelligent workflows because I think -- so you can have workflows. They have always been the harder business. Then we began to add automation to workflows, lots and lots of companies do automation, including us. We can think of the markets around business process management, workflows, et cetera. And I believe AI added to those workflows is what really drives that productivity and takes a lot of complexity out of the equation of how you integrate a workflow end-to-end, cutting across the different normal silos in a business, whether it's revenue recognition, revenue collection, quote to cash, all these processes other than HR and customer service, which I just mentioned. But your last part of the question is really important. You can think back to if you build verticals, then nobody else in that vertical is going to want to work with you, a client or another partner. So we really did really hard thinking about 1.5 years ago and said, we are better served being a horizontal provider. So I don't really want to provide financial software or marketing software, et cetera. Like I'll take an example, the reason Schlumberger was able to partner with us is because we are saying we are horizontal. We are enabling, you do the oil and gas software, you do the petrochemical suites, et cetera. And that's a much better place, I believe, for us, where we can then work across the client set with them and we get a benefit as their verticals get deployed. And you're seeing that at Palantir. We did another one with Siemens. I think these are great examples. And that's, I think, more natural for us to go do that.

Got it. I want to push on to the hybrid cloud. But just in hearing that, maybe you can give me just a quick answer. If you are departing from the vertical market strategy, there were significant investments made, Truven, Merge, et cetera. Are those divestment candidates as a result? And then secondly, there have been estimates that Watson might be $1 billion in revenue for IBM. How do we think about the relative size of Watson today?

So I'll try to address those directly. Look, we are always optimizing our portfolio. I'll just point to the past. We took out a lot of the marketing products. We divested them to HCL and to others. We also divested our X86 on-premise server business a few years ago. We divested our semiconductor manufacturing some years ago. So we are always fine-tuning and optimizing our portfolio. The lens I will use on that is we are a hybrid cloud and AI company. I defined AI as being much more platform and horizontal than vertical. So if you look upon that, the services, the products, the offerings that stay at IBM are more or less going to be aligned to that. And that was part of divesting or rather spinning out Kyndryl, which is our Managed Infrastructure business, which you can't say it has not, but it has less synergy with a pure hybrid cloud and AI company. So I'll answer the first part of the question that way. I get it. It's not completely satisfactory, but that's kind of the lens we look at it with. As we begin to look now at other areas of investment, it's going to be tied back to our hybrid cloud and AI strategy, Toni. So that's kind of how I think it's -- if that's a fair answer to your question.

And the second part, the $1 billion, is that a rough ballpark sense of the size of Watson in your AI business today?

No. So the reason why I'm hesitating to answer that question, we can't measure it directly. So the mainframe has now an AI sub-processor on the main processor. We're using AI to control every aspect of how do we manage our IT operations. We are using AI to pretty much run every process inside IBM. We use AI, I'm going to say, pretty much in every product, including how we move data, including how we transform data, how does the database optimize itself, how do I do application modernization. And to give you an example, so the bulk of the GBS business, the vast majority of that $16 billion and change business, does application development, maintenance, optimization for our clients. We have a massive library of prior projects we have done. We have an AI tool that we use internally that helps the teams determine very quickly. So as opposed to the thousands of man-days that would often take to do the upfront work, we have an AI tool, that's our own, we don't sell that, at least not yet but we use it for our people, that goes in and says, this is how much work it would be or how complicated it would be to do something. And we use that to guide ourselves on that project, to account that as AI revenue or not as AI revenue. So I actually view that AI is going to infuse everything. We talk about this, there's tens of thousands of AI projects we do. And so if I look across what fraction of GBS can I attribute to AI or what fraction of the product's going to attribute to AI, I don't believe that that's a fair measure. What it will show up in, Toni, is eventually our growth. So as you heard me say -- and it is a departure from the past. I have committed that we are now committed to growth as a fundamental part of our strategy. It's not about optimizing other financial measures. And so as we turn towards mid-single-digit growth, then that's a fair question to ask, how much of that can you attribute to AI and how much of that can you attribute to hybrid cloud. And I'll look at you and smile and say, you can attribute 110% of the growth to each of those, but there is a complete overlap.

Of course. Got it. Well, that's a good segue into hybrid cloud. So I mean, again, maybe definitionally, you can help investors. I get this question a lot. IBM has said it's done $26 billion in cloud revenue over the last 12 months, growing at 18%. But if you went to IDC or Gartner, they would say, well, look, cloud, we define it as SaaS, we define it as Infrastructure as a Service, and we define it as Platform as a Service. And IBM's revenues on that basis are lower, considerably lower, in the single-digit billions. So what is in IBM's definition of that $26 billion? And why is it so different from the number that we might get from IDC or Gartner, call it, $4 billion or $5 billion? And then very briefly, is hybrid cloud really about Red Hat and multicloud? Or how do you view it for IBM? So one, what's in it? And then two, is this really Red Hat and multicloud?

Yes. Actually, Toni, I think the answers to both those questions are the same. So if you would indulge me for a minute, let me begin with what I believe is our client environment, and that will allow me then to say what is hybrid cloud. So if I take one of our typical clients, they might run some of their workload on Azure. They might run some of their workload on Amazon Web Services or AWS. They will run and buy 6 different as a Service properties, from Salesforce to Adobe, to Dropbox, to Box, to Slack, maybe Microsoft Dynamics, and we can go on and on. Then they will also run some of their infrastructure in their own data centers. As an example, a banking CEO that I spoke with recently, I'm not going to name the person, talked about, look, I am going to use all the properties I mentioned. But from a core retail banking, I'm going to run it in my own data centers because I need my own people to know and completely know that technology, so if anything ever happens, they can fix it, they can recover it. And I look them in the eye, [ nod, know all the dependents ] for that core piece. You could, in that case, turn around and say, 40%, maybe, maybe 60%, somewhere in that range, depending on whether you account dollars or importance or volume, is sitting in private, aka their own data centers. 60% is maybe sitting across the different clouds and as a Service properties. I think that's reality for many of our clients. So it's multiple public clouds, aka multicloud. It's multiple SaaS. And it is private. If that's the environment, that's what I call hybrid cloud. So now to go a bit further, what do I mean by our hybrid cloud platform and Red Hat and what's in our numbers. So if that's hybrid cloud, some of the workload, they're completely fine to say, look, you can go to a single public cloud, where you can go completely locked into those APIs, those services, those innovations that come there, but then it is confined to that public cloud. For some others, they're going to say, whether it's because of regulatory reasons or it's because of brand concerns or it's because of latency, they're going to say, look, I want the ability to take it from private to public or public to private. I need the ability to run the same workload on 2 different public clouds because the regulator might demand that. So for that portion, and that's what we estimate as the $1 trillion TAM because the first environment I described is all of IT. That's $3 trillion. So what -- where people want the ability to have some flexibility, not on a day or not in an hour, but maybe over years, maybe over 6 months, to make a change in direction, that's what I'm calling is the hybrid cloud market explicitly. Obviously, regulated entities are going to have more of an affinity for it, not exclusively, but more. And unregulated entities will have a bit less of an affinity for it. So that is hybrid cloud. It is inclusive of the capabilities that you put in there, aka products, and it is inclusive of the services you do to help people migrate or run in those specific environments. That's in our cloud number of the $26 billion. So it's a fraction of the software. It's a fraction of GBS. It's a fraction of[ IES ] as long as it's still in here. And we give those numbers for each of those businesses. Now that you say, is it only Red Hat? Well, you do need, I believe, fundamentally a core platform. So Red Hat gives me the core platform that runs across the different public clouds. And you see that in the combination of RHEL, or Red Hat Linux, as well as OpenShift. It runs across those public clouds. It runs across private also. And that gives you the ability to have a technology answer and a platform that goes across those. But that's not all that's in that cloud number. Then there are products, for example, our Cloud Pak for Data, that runs on top of OpenShift that's going to be in that number. There is all the GBS book of business, close to $6 billion now of doing work on cloud that sits in there. By the way, 1/3 of the GBS signings on cloud are related to Red Hat, but there's another 2/3 that are not. So it is wider than Red Hat, albeit Red Hat is the fastest-growing and a really important part of that puzzle for us. So hopefully, that made sense, to explain why we think of hybrid cloud. Our point of view is a little bit different because many people equate cloud to public cloud only. And I'll take a great client of ours, a different example than a bank, a telecom carrier like Verizon. When they use the word "cloud," they mean their own cloud because they have hundreds of thousands of servers that run their network. I think it's unlikely they would take that to any single public cloud. So is that a cloud or is that not a cloud as they go from 4G to 5G? They would look at you very strongly in the eye and say, that's a cloud because we're using virtualization, we're using containers, we flex the workload, we can move the workload around amongst our own different data centers, et cetera. And that's the world of hybrid that we assert is the right way to look at cloud. And I think as edge comes on as more and more people are going to move workloads closer to the edge, whether it's warehouses, big retail stores, bank branches, telecom, 5G, I think we're going to see that there is a movement back where there is a distribution between the big centers and the edge. Sorry, Toni, long answer to your question, but you let me go in my passion of explaining what this world is and why we do [ hold it ] slightly different. It's not an oppositional view. Actually, it builds on what the hyperscalers do. They're great partners for us, but it's a slightly different definition, and we are trying to solve a client problem that is, I think, additive and incremental to what they do.

Right. So let me follow up with 2 questions on that. So is IBM's differentiation through Red Hat truly multicloud, a platform by which I can migrate a workload or an application from one cloud to another, as you suggested, between private cloud and public cloud or various public clouds. Is that the key defining differentiation for IBM going forward? And then secondly, at a high level because we have a lot of portfolio managers who listen, can you discuss what Cloud Paks are and how they relate to that?

Yes. So let me unpack the first part of the question, and I'll try to keep it to a high level and give an economic thesis of how we look at this. Absolutely, the Red Hat platform is a key part of that differentiation. It gives us -- we can use many phrases. The old phrase, I remember from 20 years ago, and I know, Toni, you will but not all the audience might, is the right one [ to run anywhere ] model. That was the world of Java 20 years ago, and I think it proved to be very successful because the applications weren't confined to any given server at that point. Now you say if I want to go from private to public, or actually, I might [ run in ] public today, but might want to go to public -- private today, but go to public tomorrow, or I might say, I'll do my depth test on public, but want to run my production on private, we give you a platform that lets you do that. And then it's really important that it runs on multiple clouds, but that's why you see the Red Hat portfolio is in the cloud catalogs at both Amazon and Azure. Then I say, but a certain fraction of my clients, maybe not all, but we can say maybe 30%, maybe 50%, will buy more capabilities from us. They will buy the ability to move data. They might buy the ability to run their Java workloads. They might buy the ability to do IT operations with even less labor and less complexity than before. So then that gives us a multiplier. If a client does do that, that gives you $3 to $5 more than just having the Red Hat platform. But not all of them will do that. So I wouldn't put a multiplier of 5. You got to do the multiplier of how often do we succeed on that as opposed to that. There are some other clients that are going to turn around and say, I really like the journey. I'm going to go to different clouds. I need to modernize my applications. As I do that, I need to think about where I get value. So there's a services multiplier of $6 to $8, if there is a core view on the platform itself. So then you say, what fraction do you get there. So really, like $1 on the platform win turns into $3 to $5 as you look through that virtual flywheel. By the way, it's not one way. If they use our services, GBS now has the largest Red Hat practice. Actually, $2 billion of signings -- inception-to-date signings on the Red Hat platform. That means the platform gets deployed faster. It gets consumed faster. That means more of it will get consumed. As you do [ reference cells ], then you get more of a multiplier. So it's kind of a flywheel. You get your platform, you put more software capabilities, that makes it more attractive. You also work with others to bring their capabilities. You get our services and other people's, that gets the whole thing deployed faster. More gets consumed, more innovation, more experiential learning, more skills out there and sort of you get into this flywheel effect that's going on. So that's the first part of the answer, Toni, in terms of is it just Red Hat or a lot more. Remind me of the second part of your question?

Just simplistically, what Cloud Paks are and how they fit in.

Yes. So we want all of our software to fit the same thesis as I described for Red Hat, run it on all clouds, including on private and including on IBM Cloud. So Cloud Pak is a way to take a lot more innovation as well as some of our existing capabilities and run them on top of OpenShift to go out. I'll give an example from the Cloud Pak for Data, which is 1 of the 6 we had. So Cloud Pak for Data says, look, I'm going to take a lot of open source and IBM technologies. I'm going to let you discover all your data in the enterprise, including that which is on your version on your public -- your estate in the public clouds. I begin to try and understand that data automatically using AI. So I would give you a catalog so that you know where your data is. For a lot of things, you don't even need to move your data. You can just do it in a federated way. For other things, you might wish to move your data. And then other people take advantage of the Cloud Pak for Data because that gives them the access to their environment. Example with Palantir, Palantir is running on top of the Cloud Pak for Data for some of their markets because that gives them an access to how we do all this and they don't need to rebuild all that scaffolding. And so you can think about that's the data side. There is -- and by the way, the AI capabilities come built in. There's another one for automation, where you begin to take -- think RPA nowadays or BPM or combined and how do you automate. There is one for IT operations, which actually builds on the automation one. So you can imagine that, that is what Cloud Paks are. So as opposed to hundreds of products, you're coming down to, really, there is 6 Cloud Paks today. I could see a way up into maybe 10, maybe 15. I don't really see us doing hundreds and thousands. You put them all as bundled capabilities, all of which then work together inside these Cloud Paks. So it's a lot more than just technology. You have to get into what's the ease of use, what's the experience, what's the pricing models because you got to fit the pricing models on these different environments, et cetera.