America used to run on IBM. It was the backbone of business. But after getting that jingle stuck in the heads of millions of Americans in the late 1980s, Big Blue lost its swagger.
IBM is getting a lot of attention today. More now on that big miss for IBM on both earnings and revenue.
It's been a rough go. There's no doubt about it. IBM spent much of the 2010s in the doldrums, but has made something of a comeback in the past 5 years. That's thanks to a lot of the success in its hybrid cloud business. They've also leaned heavily into consulting services, all of which has led to a surge in the company's share price. That's under the leadership of CEO Arvind Krishna. He took the reins in 2020. And now IBM's looking to the future with quantum computing, something I saw firsthand earlier this summer. I'm standing outside the Thomas J. Watson IBM research center in Yorktown Heights, New York. Believe it or not, this is probably the leading lab in the world for quantum computing research. But betting on the next big thing hasn't always worked out for IBM. Just look at AI. In 2011, IBM's Watson AI famously beat Ken Jennings on Jeopardy.
Usually when I play Jeopardy, it's just for the fun of it. And this time, I feel like it's all part of some vast sociotechnological experiment.
What is Jericho? Correct. It was kind of downhill from there. It felt like IBM had lost the plot. Now we ask Arvind Krishna how IBM is going to recapture the future through quantum computing.
Arvind, welcome. I'm excited to have you here in part because I was just at your Thomas J. Watson Research Center in Yorktown Heights. Very impressive, built in the 60s. Looks like the set of a Kubrick film. And I got to spend time with your head of quantum hardware, Jerry Chow. The technology here is really complicated. I spent three hours with Jerry and basically my summary of how your quantum computing works is that it's magic. So I want to skip over the hour we could spend describing how this works and go straight to applications. What applications of your quantum computing technology are you the most excited about?
I think the reason we are so excited about quantum is its ability to solve problems that normal computers cannot, and actually I'll make a stronger statement, will not solve. So if we want to understand how materials really work as an example, could we design a molecule that's better for carbon sequestration? Could we come up with a way to fix nitrogen so that we can increase food production and quality in the world because that's fertilizers to make it simple. Could we come up with a coating to reduce corrosion in underwater pipes so oil and gas don't ever leak into the ocean? Those are the kinds of problems that I'm super excited about.
Right. Because you're talking about doing simulations of quantum effects. And funny enough, you need a quantum computer to simulate the way atoms behave at that level.
Richard Feynman's quote, if you want to simulate nature, you need a quantum computer. He added some more colorful language, but for the podcast, I'll stop there. And to give a bit of intuition, when you begin to look at these materials or even simple molecules, you very quickly get into this 30, 40, 100 electrons and in order to compute them, that's impossible. If you try to simulate them on a normal computer, you begin to need the amount of memory that exceeds that that is even possible because 2 to the 200 is a number we just can't fathom in terms of the amount of state that is needed. Quantum computers will be able to do that shortly.
So a bunch of other tech companies are also eyeing quantum, right? You've got Google, you've got Microsoft, which have reported some pretty exciting breakthroughs in the past year or so. Why is IBM the company that's finally going to crack this technology?
So, I'm going to give you a couple of statements which are somewhat dissonant. Number one, I love the fact, really I really love the fact that a few dozen or more others are going after this technology. Why? A, it helps to make a market because if it's just us, you'll begin to question why should I believe you? Well, if others aren't trying it, is it really of any value? So, it actually is incredibly useful from a client, government as well as I'll say from a media perspective that there's many others because I think that defines a market as opposed to not.
The race for the future. Right. Now it comes to so why would we be a winner of the race?
And you know Tim, I'm saying a winner, not the only winner.
I got it. I want my technology teams to be super excited to be the only winner. But as long as we are a winner, I'm good. Okay. So now why would we do that? I think a lot of people are working on incredibly exciting, and we haven't yet used the term qubit, but let's think of that as the fundamental building block of a quantum computer, not the only, but a fundamental building block. People are working on really exciting qubit technologies. I think that's great because it opens up many options and science usually advances by people standing on each other's shoulders. So actually I think that's wonderful. But now you need to connect thousands of qubits together because to compute you need to have, let's call it, signals flow from one to the other. Then it's not enough to have those connected. You need to be able to read and write to it. Then you need to be able to have it function all day, all week, and all month without needing a team of PhDs to come and tune it between every single run. Because if that's the case, that's not really a computer. And that's really the game though, ultimately, reliability.
Yeah. Couple of different answers to what you're describing. Number one, we need the computer to function. Then there is the whole question of because you're trying to operate in very, very tiny energy states. That's the point of the cooling. Cool it down so that the normal thermal effects can be taken out of the picture because they're not going to happen at these super cold temperatures.
Well, Chris, to be candid, that's a pretty straightforward problem.
Yeah, it sounds interesting. No big deal. Just colder than deep space.
It sounds like one of those, you know, really, you said Kubrick, I'll say Arthur C. Clarke. Same difference there.
Because I'm thinking and channeling 2001: A Space Odyssey. Open the pod bay doors. Hell, I'm sorry, Dave. I'm afraid I can't do that. And it begins to sound like some kind of industrial or maybe post-apocalyptic scenario, but that is actually pretty standard technology to me. It's arcane. It's not ultra cheap, but neither is it ultra expensive. So, let's put aside the cooling and all of the things. The question becomes, can you take what is inherently going to be unreliable? I make that very strong statement at a single qubit level and stitch it together using, I'll use the word, coding techniques so that what is there is a lot more resilient to errors. All that said, it's still going to be resilient to errors for maybe 100 milliseconds to a second and then you have to start again. So how much computation can you get done in that amount of time is I think the singular measurement for quantum computers.
Wow. So a lot of math there, a lot of science, a lot of science fiction references. We got a lot there. How more, maybe when do you realistically expect to see a return on IBM's massive investment in this area? When do you expect this to be commercial? When do you expect this to start happening?
I would tell you that we are looking at 2029, 2030. So that is 4 to 5 years out for that return that you're talking about. Now that's the beginning of the return and then you'll begin to see people beginning to use these. Then I think a year or two after there'll be even more and then a year or two after there'll be even more. We have seen what happens when technology begins to work and to scale. The first few years are remarkably exciting about adoption.
For anybody listening, in case they haven't guessed, you are an engineer. You are the first engineer to run an IBM ever. How is that helping you navigate this time? Because this is a time of, I would say this is a time of more rapid change for IBM than at any point in the entire history of me watching your company.
I kind of call it, I'm an unabashed technologist and geek at heart. I love trying to figure out and hear explanations of how these things work because that I think, Christopher, answers the question you're asking. So, how does it help me when you have these really intrinsic and arcane technologies that kind of seem like science fiction? Having a bit of the background, it helps me understand kind of two things. One, it is going to take time. So, we're still four years out, but two, when we get there, I now know how to scale really, really quickly to go get it into lots of people's hands. So, it helps make those decisions. It also makes the conversation much, much faster with the teams that work on it because I don't have to go through a layer of intermediaries or trying to translate it into different language. I mean they can talk to me how they want to. I may understand maybe half of it but that's better than understanding only 1% of it.
And I just want to flag what you've already mentioned which is that 2029 is when you're going to really start scaling your quantum computing efforts with Starling. Right.
Correct Christopher but for the following reasons. We got two steps to do between now and then. Number one, you got to get these systems much bigger. Number two, you've got to prove that the error correction works.
Clearly, you've got a lot of momentum. You were early to quantum. You're kind of the bell of the ball. But how do you stay ahead? How do you maintain this momentum?
To maintain the momentum, we have to grow a very large ecosystem on quantum. So the use cases multiply and we might perhaps collect 10% of the value, 90% of the value should go to the people and the consumers of the technology not to us. So educating people, teaching them how to use quantum, unlocking their imagination is a big part of what we have to do. And I'm honest about the 10/90 trade-off. I think that that is where it has to go otherwise the technology doesn't really scale.
In the today though, I think investors are very excited about the potential for AI and where they see IBM playing in that space. That's the hot technology of the moment in the valley and in a lot of investors' minds since you became CEO in early 2020. This IBM stock has just been on a tear, right? I think your stock in a lot of ways is trading with the kind of premium that you see with the AI heavyweights out there. And I wonder, I want to talk a little bit about how you kind of see AI in the business and how you're balancing kind of that with your traditional role because I think one of the things that IBM has received a lot of praise for under your tenure is really clarifying its consulting business, right? You've worked to make IBM really a facilitator for your clients, making this tech transformation not necessarily steering them into IBM products per se but growing a consulting business and I wonder with AI, doesn't AI potentially disrupt your consulting business as it changes work in general? Do you go to bed afraid at night or excited at night?
Yeah, I am actually excited. I lose no sleep over what AI will do to our consulting teams. Why do I say that? Do I fully believe that, and we can debate is it a third, is it half, is it two-thirds of what people do in consulting that work will be replaced by AI and the collection, reasoning AI agents, all of that. I'm actually firmly convinced of that. We are running hard towards making that happen. So you then say wait if half the work happens don't you lose half your consulting business and I go to no, the exact opposite happens. If we run towards it that means we are more productive in giving people consulting help. If you are more productive, history has shown that the more productive company gains market share because if you offer higher quality things at a lower unit cost you tend to take market share and if you take market share you'll actually win more customers. So that is why I believe that that is the route we need to go and go at it hard and that is how we are going to go win for AI in and for big company adoption or just company adoption in general.
It seems like we're still in this area, this era, mostly experimental, trying to figure out how this might be useful. We're seeing some early, early things. But I'm curious what you're hearing from customers and how that conversation is changing on what they want from AI in the now and in the next few years in the tangible.
So let's acknowledge the vast majority, as in over 80% of the clients we talk to, will acknowledge they're still in the era of AI experimentation. You know, typical conversation goes like, 'How many AI POCs are you doing?' I'm using proof of concepts as the jargon for experiments. 'Oh, I'm doing a hundred. I'm so proud of it. I got every function doing a little bit.' Okay. How many of them do you think are going to scale? And they kind of look at the shoes. Like one of them told me, 'Well, I think maybe three.' I said like, so is the flaw in your design of how you're using AI? And then they begin to acknowledge and that is the push now on the ROI point. ROI is only achieved when a technology begins to give value back to the business. Doing the experiment has no value in itself except to prove feasibility. So they've all begun to ask the question now how do I scale and how do I get that value? That means I need common tool sets across the different experiments. I need to think about some of the questions up front like what is the life cycle of a model? What is the governance? How do I make sure that I am correct about privacy and all the other aspects that we always need when we are dealing with data and with critical functions. And so if I was to use a sports analogy, I would tell you AI is in the first innings. So players are on the field. You know who's who's going to play, but it's still early, early to see how the game works out and how it goes along.
You know, listening to you talk about quantum. I mean, clearly you're excited about it, right? But I also feel like you're tempering your enthusiasm a little bit. The old IBM perhaps, and I think of kind of the AI IBM talking about Watson.
