# Masters of Scale: IBM’s $10 billion bet on what comes after AI

- **Show:** Pioneers of AI
- **Published:** 29 Jun 2026
- **Source:** https://prfx.byspotify.com/e/dts.podtrac.com/redirect.mp3/tracking.swap.fm/track/7G0M4p2bOLP0mP50O3CD/pscrb.fm/rss/p/rss.art19.com/episodes/a764f5ef-2643-4ff7-82f8-ae105d2063e5.mp3?rss_browser=BAhJIhlwb2RjYXN0Lm9wZW5uYXNoLmNvbQY6BkVU--4dcea1dd70768d62b2466884c2f8513b5f848fbf
- **Transcript:** https://podcasts.opennash.com/shows/pioneers-of-ai/masters-of-scale-ibms-10-billion-bet-on-what-comes-after-ai-c998ad/
- **Provider:** OpenNash Podcasts (machine-generated)

## In brief

IBM CEO Arvind Krishna argues that AI's real value for businesses lies not in ever-larger foundation models but in using the right-sized tool for each job. He predicts foundation models will become commodities with little switching cost, while token prices rise, pushing companies toward smaller, cheaper, on-premise models, and he says quantum computing is the next major technology wave.

The central tension is between AI's genuine productivity payoff and the shaky economics of the AI build-out. Krishna says IBM's own AI savings now far exceed its costs, yet he doubts the math behind trillions in data center spending. He ties this to risk: companies that avoid risk decline, so leaders must reward bets with roughly 50% odds of success.

## Key insights

1. **Foundation models will likely become commodities** Krishna says switching costs between models are already low and will fall further, while GPU pricing has doubled in six months. He expects companies to move toward smaller, cheaper models that cost about one hundredth as much to run.
2. **Start with a few AI projects at scale** Krishna advises picking three to five use cases and doing them at scale rather than a hundred experiments. This teaches change management and data organization, which then lets a firm tackle ten, then twenty more.
3. **IBM's AI savings overtook costs after year one** Krishna says IBM spent more than it saved in the first six to twelve months, then saw about ten times return by year two. He expects over $5 billion in savings by year four against a 2022 spending baseline.
4. **Cheaper software development can increase hiring** Krishna says IBM developers are 40% more productive, yet it tripled entry-level hiring because cheaper development makes new products affordable. He still expects about 30% fewer people in back-office roles like compliance and procurement.
5. **Quantum has moved from toy problems to proteins** Krishna says IBM's quantum systems went from simulating a five-atom molecule in summer 2025 to 12,000 atoms in April, reaching a piece of the protein trypsin. This opens a path to understanding how drug molecules might bind to proteins.

## Questions this episode answers

### Why does IBM's CEO think AI models will become commodities?

Arvind Krishna says there is very little cost to switch from one foundation model to another, like iron or other commodities. He expects that to happen within one to three years.

### What does Arvind Krishna mean by using an eighteen-wheeler for everything?

He means companies use huge, expensive AI models for tasks that smaller, cheaper models could handle, like driving kids to school in a big truck. He predicts firms will shift to right-sized models within about 24 months.

### Is AI actually saving IBM money?

Krishna says IBM spent more than it saved in the first six to twelve months, but by year two it earned about ten times its spending. He expects over $5 billion in savings by year four compared with 2022 spending.

### Will AI cause big job losses, according to IBM's CEO?

Krishna expects about 30% fewer people in operations roles like compliance, accounts payable and procurement within a few years. He argues net demand for jobs will rise, noting IBM tripled entry-level hiring this year.

### Why is IBM investing $10 billion in quantum computing?

Krishna says IBM believes quantum can give outsized returns to those who get ahead of the curve, and that recent progress, like simulating a 12,000-atom molecule, shows it is becoming useful. He also points to the U.S. government agreeing to invest as a sign of confidence.

## Facts

- **Arvind Krishna** [0:10:03]: "Underlying GPU pricing, which is what all of these things run on, has doubled in the last six months on a per hour basis."
- **Arvind Krishna** [0:06:48]: "when we uh won Jeopardy with Watson, I think it woke the world up because I think for the first time in a long time, AI started doing something people thought it could not do."
- **Arvind Krishna** [0:07:38]: "we wanted to make a monolithic application. Mistake number one, mistake number two, we picked a domain, which is perhaps the hardest of them all, which is health."
- **Arvind Krishna** [0:05:24]: "I would say public cloud is one of them, candidly, that we missed. I would actually turn around and say, despite inventing the IBM PC, client server was another one that we missed."
- **Bob Safian** [0:14:46]: "you've, you know, unlocked four, $4.5 billion of efficiency from AI"
- **Arvind Krishna** [0:19:54]: "Our software developers are probably 40% more productive today than they were two years ago."
- **Arvind Krishna** [0:19:54]: "We actually tripled our college level entry hiring this year, tripled compared to last year."
- **Arvind Krishna** [0:20:51]: "I would not be surprised if about 30% of the total headcount in those areas is not needed within a few years."
- **Arvind Krishna** [0:24:19]: "there's 125 gigawatts of AI data centers that are going to come online in the next two to three years, that's the eight to 12 trillion of capex in total"
- **Arvind Krishna** [0:24:19]: "because that would imply close to $1 1 trillion of profit, which is, in the best case, that means 4 trillion of more revenue."
- **Arvind Krishna** [0:27:19]: "I remember in 1993, there was, I think it was Time magazine where they were showing a mainframe dressed up as a dinosaur, and it was called the death of the mainframe."
- **Arvind Krishna** [0:26:13]: "A mainframe is really useful when you have one piece of work that has an incredible volume. Example airline reservations."
- **Bob Safian** [0:28:21]: "IBM recently announced 5 billion dollar initiative called Project Lightwell to identify and fix AI vulnerabilities in the open source world."
- **Bob Safian** [0:28:21]: "that was reportedly triggered by Anthropic's UM release of Mythos."
- **Arvind Krishna** [0:28:43]: "Mythos didn't find anything that other models didn't and couldn't find."
- **Arvind Krishna** [0:31:24]: "So nation states have been doing this for decades, but you would say three or four nation states were capable of doing it."
- **Bob Safian** [0:32:18]: "You're partnered with the U.S. government on a new quantum foundry, investing $10 billion in a large-scale commercial quantum computer."
- **Arvind Krishna** [0:34:31]: "Last, uh, winter, November, December, they could do 300 atom molecule."
- **Arvind Krishna** [0:34:31]: "In April, they did 12,000 atoms. They're now getting into the protein realm."
- **Arvind Krishna** [0:35:05]: "So this was a piece of a protein called trypsin."
- **Arvind Krishna** [0:36:51]: "Two, we were about 20% software in 2019, we are now about 45%."
- **Arvind Krishna** [0:41:43]: "We're probably on our fourth try in my tenure, but we're not gonna give up it."
- **Arvind Krishna** [0:13:03]: "If you look at corporate uh performance, actually, it tends to be a twenty-eighty rule, more of a power law than the K is really more of a fifty-fifty."

## Transcript

**[0:00:01] Ad read:** Stripe is built for whatever comes next because AI isn't just changing business, it's evolving fraud. Stripe Radar detects fraud with the fast scale of Stripe's 1.9 trillion dollar annual transaction volume. In fact, when a card is used, there's 92% chance the Stripe network has seen it before. That means emerging fraud patterns are detected sooner, blocking bad actors, welcoming good customers. Learn more about fighting fraud with Stripe at stripe dot com slash radar.

**[0:00:35] Bob Safian:** The business world is moving faster than ever, and when change hits, we need to learn in real time. On Rapid Response, you'll hear candid conversations with CEOs and leaders making tough calls about AI, human talent, responsibility, and the bottom line, how they navigate uncertainty, pressure, and high-stakes moments. I'm Bob Safian, and former editor-in-chief of Fast Company, and I'll be your host as each episode breaks down what you need to know right now. You can find Rapid Response wherever you get your podcasts.

**[0:01:11] Ad read:** On Masters of Scale, iconic leaders reveal how they've beaten the odds.

**[0:01:24] Speaker 6:** Asking really strong questions is a superpower.

**[0:01:27] Ad read:** You want to show up with something radically different, and how they've grown companies to incredible heights.

**[0:01:33] Speaker 5:** The greatest rewards always come from the greatest risks.

**[0:01:36] Ad read:** Let's hit the gas. Airbnb, Zillow, Microsoft, Liquid Death, and more. Hear from the founders who've changed the game. It's anything but business as usual. Find Masters of Scale on Apple Podcasts, Spotify, YouTube, or wherever else you get podcasts.

**[0:02:02] Arvind Krishna:** I'll say something provocative. I think foundation models are going to become commodities. I think that right now, the token price on all of these is going to go way up. It just has to to justify the capital investments.

**[0:02:17] Bob Safian:** That's Arvind Krishna, CEO of IBM, and he has a strong metaphor for current AI systems that are just not the right size for all uses.

**[0:02:27] Arvind Krishna:** I guess you could, for those in the suburbs, take your kids to school in an eighteen-wheeler every morning. You could go milk shopping in an eighteen-wheeler. Then you'd ask yourself, is it really the most effective vehicle for that? I think right now we're using the eighteen-wheeler for everything.

**[0:02:44] Bob Safian:** This is Masters of Scale, I'm Bob Safian and your host. IBM is playing a distinctive role in the AI race, not building AI models, but betting on how best to use them and on what comes after them. In this conversation recorded in front of a live audience as part of New York Tech Week at IBM's Manhattan HQ, we dig into why Arvind thinks most enterprises are using an eighteen-wheeler for every task. Plus, what kind of risk-taking businesses need to take right now, how to think about cost versus benefits when implementing AI, IBM's big bet on quantum computing, and much, much more.

**[0:03:26] Bob Safian:** Please welcome to the stage IBM chairman and CEO Arvind Krishna. Arvind, first of all, um, thank you for for hosting us at your house. We're together as part of New York Tech Week. Uh, IBM is a New York institution, it is a global institution. Um, as a company, it's had to continually reinvent itself, you know, from from mainframes to PCs and consulting and cloud, and now AI and uh, and on the cusp quantum computing. How do you think about staying fresh as as tech as tech moves?

**[0:04:08] Bob Safian:** And are there things about IBM's legacy that are an advantage versus what isn't an advantage?

**[0:04:14] Arvind Krishna:** Yeah, look, um, the advantage is, uh, client intimacy, knowing your clients. The advantages are around trust. Uh, I don't think we have knowingly ever done anything wrong with a client's IP or data or people. Those are advantages. I think our people are incredibly technically adept. I would say they're experts in the areas where they have spent time and energy, so those are all the advantages. Now, technology keeps changing, and I'll be the first to acknowledge. Sometimes we are really good at predicting where it's going to go, and we kind of get ahead of the wave and do that.

**[0:04:58] Arvind Krishna:** You mentioned a few of those. Uh, the mainframe certainly, but this is now sixty years ago was one of them. Uh, I would say the IBM PC may have been another one of them. I think embracing Java in the internet era was another great one. And then you sometimes miss them. It's not for actually lack of knowledge. You miss them because the business model doesn't align. You don't quite know how to get the investments and the returns to come together.

**[0:05:24] Arvind Krishna:** I would say public cloud is one of them, candidly, that we missed. I would actually turn around and say, despite inventing the IBM PC, client server was another one that we missed. So I think it comes into you could probably miss some.

**[0:05:39] Bob Safian:** And you kind of know you're gonna miss some sometimes, or

**[0:05:41] Arvind Krishna:** And you know, you're gonna. I think because if you don't take risk, you're never gonna succeed. So part of it is, hey, I'm getting a lot from that one. I kind of want to keep my focus there. And kind of if you're not two, three years ahead of the wave, you're gonna miss the wave. But uh, the whole point then is, can you do enough of those where you can be with the wave as opposed to way behind?

**[0:06:03] Arvind Krishna:** Because it moves so fast nowadays that if you're two, three years behind, you're not gonna catch up. And so right now we are very focused on hybrid cloud, that is sort of our answer to, uh, this, uh, the cloud movement and AI, where I think that our play is going to be much more of similar to a hybrid play. How do we help our enterprise clients take advantage of it fully?

**[0:06:24] Bob Safian:** I mean, the the the first-mover advantage is this sort of a catchphrase in in in tech land. And w- I was thinking about, um, you know, AI and IBM Watson, and you were sort of ahead in some ways, but you didn't maybe make the splash that you wanted. W- was that a was that a missed moment, or is it more that, you know, you were too early, or the tech wasn't quite ready, or

**[0:06:48] Arvind Krishna:** There's always all those things. I think when we uh won Jeopardy with Watson, I think it woke the world up because I think for the first time in a long time, AI started doing something people thought it could not do. Okay, unfortunately, it woke the world up completely, in the sense that uh, a number of other companies started investing very heavily. Now we had an advantage. We might have been able to succeed, but I'm not putting on uh, 2020 vision looking backwards.

**[0:07:21] Arvind Krishna:** The mistakes we made were actually much more of a strategic nature. As opposed to creating building blocks, we wanted to create solutions and verticals. That, I think, is a mistake, as technology shows. In the beginning, it's

**[0:07:36] Bob Safian:** you went too quickly to the application?

**[0:07:38] Arvind Krishna:** we went too quickly, and we wanted to make a monolithic application. Mistake number one, mistake number two, we picked a domain, which is perhaps the hardest of them all, which is health. Mistake number three, how many IBMers sell to doctors, and how many of us IBMers deal with the FDA? Like none. So you picked the wrong solution set in an industry you know nothing about, and with a customer you know nothing about. Other than that, it was pretty good.

**[0:08:12] Bob Safian:** So so as you, um, I mean, as you think about IBM today and its r role in the AI ecosystem, like, what is it? I mean, you're not, you're not trying to be OpenAI or Anthropic, you're not trying to be Google or Microsoft. Like, are you ahead? are you behind? You know, how do you think about all of that?

**[0:08:31] Arvind Krishna:** So we are not going to be a hyperscaler, which was two of the four you mentioned, and we are not a foundation model provider. I'll say something provocative. I think foundation models are going to become commodities. By the way, not that far out. Is it a year? is it two years? is it three years? Commodities doesn't mean that they don't have value. Go to the commodity. So is iron. So commodities means that there is very little switching cost to go from one to the other.

**[0:09:04] Arvind Krishna:** Second, I think that right now, the token price on all of these is going to go way up. It just has to to justify the capital investments. You put those two together, and there's going to be a huge motivation then from everybody who's using them to say, I need to optimize, I need to e-use each one, but in the most economic way possible. Our role is to a let our enterprise clients do that, to do it in a way that is safe.

**[0:09:33] Arvind Krishna:** And right now, there is very little demand for on-premise or smaller models, which are effectively then one hundredth of the cost to run. So I'll use the analogy in a very crude way. In the end, a automobile, which I'll include trucks into it, is an automobile at some gross level. If you step back, I guess you could, for those in the suburbs, take your kids to school in an eighteen-wheeler every morning. You could go milk shopping in an eighteen-wheeler.

**[0:10:03] Arvind Krishna:** Then you'd ask yourself, is it really the most effective vehicle for that? But if you're moving homes, which you do every seven years on average in the country, it is the most effective vehicle for that. I think right now we're using the eighteen-wheeler for everything. And this is the transition that I'll predict will happen within twenty four months. I'm not sure it'll happen within twelve months. Underlying GPU pricing, which is what all of these things run on, has doubled in the last six months on a per hour basis.

**[0:10:33] Bob Safian:** It's getting more expensive to use these tools.

**[0:10:36] Arvind Krishna:** And right now, when you're pre-public, it's okay to lose money because you're gearing towards number of customers. I think you've seen this before, right? It used to be called eyeballs, and then it suddenly became the economics are important. So I think we're right, maybe a year from that point.

**[0:10:51] Bob Safian:** I mean, you you said something at the IBM Think event a few weeks ago. You said it's day zero of the AI revolution, and and I think for a lot of folks, it feels further along than that. I mean, you've got, you know, 1 trillion dollar AI companies. A a lot of business leaders worry that they're falling behind. I I does day zero mean like, you're not too far behind, you don't have to rush too much, or where do what do you mean by that?

**[0:11:18] Arvind Krishna:** Yeah, so first, let me be clear, because I can sound a little bit cynical about the economics and and I actually am. That said, I think AI is an incredible productivity tool. I think those who don't take advantage of it will be perpetually disadvantaged compared to those who do. So let me begin by saying that it's going to optimize how you market, it's going to optimize how you write code, it's going to optimize enterprise operations, it's going to optimize how you sell, it's going to optimize how you get your daily work done.

**[0:11:49] Arvind Krishna:** So there is no question about it. It's going to make a profound and deep impact on all of those things. By day zero, I mean, it's time to sit down, take it seriously. You're not in the experimentation phase. This is not like you're in high school. Day zero, the race is about to start. Put yourself in the blocks and start sprinting. But by that, I mean, take three, four, five things, not a hundred, and learn how to do them at scale, because that'll teach you how do you get uh all your change management done?

**[0:12:25] Arvind Krishna:** How do you get your data organized? How do you really get people motivated to change a process? So do a few things at scale, learn how to do that really well. Now do ten, and then give yourself the confidence to do the next twenty.

**[0:12:39] Bob Safian:** I mean, there's, um, there's this expression that's used to talk about the economy these days that it's a K-shaped economy. You know, some households do great and and some don't. As well, sometimes I get the sense that when it comes to AI, we're sort of having K-shaped businesses. That like the tech companies, folks like you are super excited, and then there are a bunch of other companies, and and these may be clients of yours, I don't know, that are like falling behind.

**[0:13:03] Arvind Krishna:** Yeah, so, uh, unfortunately, I think corporate performance is even more of a um differentiated K. If you look at corporate uh performance, actually, it tends to be a twenty-eighty rule, more of a power law than the K is really more of a fifty-fifty. I think if I follow the economists correctly, here it's more of a twenty-eighty. And so 20% get it, they go forward, they're jumping into it, they kind of are going to get their returns, and 80% are either not getting a return or don't quite know what to do.

**[0:13:37] Arvind Krishna:** If you're in that eighty, figure out what is it that you should do. And it probably doesn't matter where you start, as long as you're starting to try to do it at scale, to make a real difference to your bottom line.

**[0:13:48] Bob Safian:** So y y what you're saying, though, is like, y you don't have to know what to do. Ju like, it's better to pick something and go than to just be like, I'm not sure what to do.

**[0:13:58] Arvind Krishna:** Like, I have a client walk up to me and say, Look, I get it that I need to do it, but I don't have the right people on my team. Can you give me a deep AI expert, somebody who's kind of done their PhD in AI? And I looked at them and I said, Actually, I recommend we give you somebody from a domain who doesn't really know the depth of AI, but who understands the difference AI could make to your domain.

**[0:14:18] Arvind Krishna:** So you don't need to know what to do, because those domain experts exist in every company. Find that twenty or 30% of them who are motivated to say, I want to learn a new way to do things. So I think curiosity, uh, willingness to adapt is more important. I think we're getting hung up on. I need to know AI like a PhD in computer science. I think that's the wrong thing because that's for the inventors of AI.

**[0:14:44] Arvind Krishna:** That's not needed for the deployers of AI.

**[0:14:46] Bob Safian:** There's also, um, you know, this idea that AI is gonna save me a lot of money. It's gonna be very efficient for me. And I think for a lot of businesses, when they start implementing, uh, those results don't necessarily come. Now, you you guys have talked about that, you've, you know, unlocked four, $4.5 billion of efficiency from AI, so you're doing something. Um, but there are also these like hidden costs, as you men as you mentioned, tokens, like, how do you balance what are the costs versus the efficiency and what you should be expecting?

**[0:15:19] Arvind Krishna:** Yeah, so this was my point of scaling. I would probably turn around and say that for our first six months to a year, we were probably spending more than we were saving. Because if you think about, you're putting a couple of hundred engineers to work at it, that's an incremental cost. The underlying infrastructure, aka the tokens, if you're doing it on public, is a added cost. There's opportunity cost also of not doing other things with these people that could have resulted in revenue.

**[0:15:48] Arvind Krishna:** That's a cost. Now, once we learn a rinse and repeat method that you're not doing it across two or three, but across ten or twenty, when you're saving $1 billion a year, well, that's a lot more than the cost of a couple of hundred people. After year two, we were definitely getting a return that was ten times compared to what we were spending. And now at, uh, year four, uh, we'll be, I think, over 5 billion from a baseline of twenty two spend.

**[0:16:15] Arvind Krishna:** So that's not an incremental five over last year. That's a incremental saving compared to our year end twenty two spending. So that is tremendous. That is more than enough to offset any extra expense.

**[0:16:28] Bob Safian:** Uh, there there was a CEO I was talking to about some of these issues. We were talking about the inexact nature of some of the outputs you get from AI, right? I don't want to call them hallucinations, but you know, the things that don't go the way you want. And he

**[0:16:41] Arvind Krishna:** It's so unlike humans, right?

**[0:16:42] Bob Safian:** so not so unlike humans. Well, what he was saying was that the money that he was saving by having his engineers use AI, that on the few cases where it was wrong, he had to spend so much trying to find what was wrong and fix it that he wasn't actually coming out ahead.

**[0:17:02] Arvind Krishna:** Yeah, so I actually think that that is an edge case of how you use AI that I think is actually wrong. I think that you should try to use AI in a case where you're not going to have to go undo six months of work or undo having spent hundreds of millions. Take customer service. If it gives a wrong answer, you got to undo one customer service answer. Then you can put all kind of evaluations and checks so AI can check itself to make sure that you're not like way off in the wild.

**[0:17:36] Arvind Krishna:** Okay, it may be slightly off, but you're not way off. So you can put checks and balances, and this is the sophistication of how you use it. So when we use it, for example, for our software developers to help them code, I don't think they realize it. We actually have checks built in to make sure that what it's suggesting is not absolutely crazy, right?

**[0:17:56] Bob Safian:** Right, but do you don't you? I I mean, as with a a human worker, you have to expect that sometimes it

**[0:18:03] Arvind Krishna:** It will go wrong.

**[0:18:03] Bob Safian:** it will go wrong.

**[0:18:04] Arvind Krishna:** So I kind of turned around. If I look at customer service, I think the stat, which would be a good one, is 85% of the time, the human people get it right, and 15% of the time. And I say, of course, humans get angry, humans get pissed off, humans will not like the tone of the person on the other end. If it's a call, humans are sometimes overconfident. I'm sure we all remember things perfectly, right?

**[0:18:25] Arvind Krishna:** I mean, you're at perfect records,

**[0:18:25] Bob Safian:** Oh, you and I do, I'm not sure everyone else

**[0:18:27] Arvind Krishna:** our spouses have never told us that we are completely in the wrong and remember it completely. So humans have all those too. I think AI is at least, if you keep it constrained to some extent, is probably 95% correct. So the evals I'm talking about is more like saying, when you think you're way off, punt it to a human. Don't try to venture into the underconfident, uh, range.

**[0:18:51] Bob Safian:** but I mean, but the AI is is always confident, right?

**[0:18:54] Arvind Krishna:** And after that, after that, one point, no, it's not actually. You'll be surprised. If you tell it, don't pretend to be confident, you will get from it that, hey, I'm not quite sure that this is as this thing. And you can put a something to check it. Who who's rewarded on actually spotting out the other's mistakes? So now you have,

**[0:19:11] Bob Safian:** So this is the this is the advantage of having multiple agents, mu multiple multiple models, multiple models working working at the at the same time.

**[0:19:20] Arvind Krishna:** just like humans. You put four eyes, we call it four eyes in software development, you put two people, one is coding, one is checking their work, just like the models, one is doing something, the other is checking its work.

**[0:19:30] Bob Safian:** And and when you? But when you have two models instead of two people, does that mean that you need fewer people? I mean, that that is one of the you know, you, you got some grief a few years back for saying, you, you know, your your back office was gonna get smaller, which is seems like it's like small change compared to the things that some other CEOs are saying right now. Do you think there's gonna be a lot of job displacement?

**[0:19:54] Arvind Krishna:** I think this is the so I'll address both parts of the question because it's a and it's not a so. Our software developers are probably 40% more productive today than they were two years ago. So I'm not saying it's one month over two years, they're 40% more productive. So you could turn around and say that means you need 40% fewer developers. We actually tripled our college level entry hiring this year, tripled compared to last year.

**[0:20:23] Arvind Krishna:** So you say, wait, that seems off. No, because this is what people are missing. If my cost of software development is going down, that means we can make products that were not economically affordable three years ago. If we can do those, we can get more revenue at an appropriate margin. So why wouldn't I get more people? Because these are value creating. Then there is the 20% I'll call it is what you need to run the operation.

**[0:20:51] Arvind Krishna:** So is it compliance? is it accounts payable? is it procurement? is it all those things? It's not going to go down to zero, but I would not be surprised if about 30% of the total headcount in those areas is not needed within a few years. That's the statement I had made, and I'm actually still consistent. Note, I just said we tripled our entry level hiring. So while there is some decrease on this side, in about 20% of the enterprise, there is a big increase in the remaining 80% of the enterprise.

**[0:21:21] Arvind Krishna:** So I actually think net will have increased demand for jobs, but there is some displacement, which is always a little bit painful.

**[0:21:29] Bob Safian:** And and always happens with new technology.

**[0:21:31] Arvind Krishna:** And always has happened with new technology.

**[0:21:34] Bob Safian:** What kind of responsibility do you feel like you have? Do you think other CEOs have helping to ameliorate the the displacement that's inevitable? You know, tech folks are very excited about the future because they're beneficiaries of it, but not everyone is a beneficiary in that way.

**[0:21:50] Arvind Krishna:** Well, actually, I think that if we can get five to ten points of productivity in every enterprise around the planet, everybody is a beneficiary. Tech may be the early, early beneficiary, but I will note everybody in tech right now is losing money on it. So we can claim, I is it going to be a long-term beneficiary or not? I think there's open questions on that. I think everybody is going to be a beneficiary. I think that in our societies, at least in the West, the responsibility of business leaders is to provide an opportunity.

**[0:22:18] Arvind Krishna:** So we want to help our people get upskilled, we want to help our people get reskilled, we want to open up that there are other opportunities, or jobs. We can't force them to do any of that. But then it's on them. Do they want to take advantage of those opportunities and step up to do it? And I would say that answer has always been about fifty-fifty. Some do, but a lot of people say, I don't want to get retrained, I want my old job.

**[0:22:44] Arvind Krishna:** Okay, I'm sorry, that's not going to happen.

**[0:22:45] Bob Safian:** And and those folks, that's ends up being the responsibility of government and society, not necessarily of the business.

**[0:22:53] Arvind Krishna:** Correct. We try to be compassionate, we don't force people out like in a day. But if over six months or nine months, they are not willing to learn the appropriate skills where they're needed, that's actually bad for the other 90% who are around them.

**[0:23:07] Bob Safian:** Still ahead, why Arvind thinks the math behind today's AI bubble just doesn't add up, the cybersecurity question, keeping his team up at night, and IBM's 10 billion dollar bet on quantum. Stay with us.

**[0:23:25] Ad read:** The companies winning right now aren't just building great AI, they're building the infrastructure to actually run a business on top of it, which is where Stripe comes in. Stripe is built for whatever comes next, with planetary scale financial infrastructure to help you power any business model. As a matter of fact, 86% of the Forbes AI fifty run on Stripe, along with industry leaders like Salesforce and Pepsi. Last year, Stripe users processed $1.9 1 trillion combined.

**[0:23:55] Ad read:** See how Stripe can power what's next for your business at stripe dot com.

**[0:24:02] Bob Safian:** Welcome back to Masters of Scale. You can find this conversation and much more on our YouTube channel, and be sure to check out the link in the show notes to subscribe to our newsletter. Are there signs of a bubble that you see in different places?

**[0:24:19] Arvind Krishna:** If I take all the verbal promises, and if we say there's 125 gigawatts of AI data centers that are going to come online in the next two to three years, that's the eight to 12 trillion of capex in total, not in one year, in total. That is where I come to. I don't see the economics of that at all, because that would imply close to $1 1 trillion of profit, which is, in the best case, that means 4 trillion of more revenue.

**[0:24:50] Arvind Krishna:** Okay, where exactly is that going to come from?

**[0:24:52] Bob Safian:** The maths doesn't doesn't add up for you.

**[0:24:54] Arvind Krishna:** That doesn't work now. Will it really, really work out well for at least half of them? Yes, some are going to thrive, but some will disappoint. So, and I don't think in a commodity world, there probably isn't space for a dozen foundation models. Is there space for three or four, probably, but since there's a dozen being run after globally, it tells you that they're not all going to work out.

**[0:25:21] Bob Safian:** So I'm gonna ask you a super basic um technology question, which, uh, which will maybe reveal something about me, but may help those in the in the room. What is the difference between a data center and a mainframe? I mean, aren't they both buildings with a lot of boxes in them?

**[0:25:39] Arvind Krishna:** So, um, first, for the ge few geeks in the room, a mainframe is actually one box, but a mainframe is designed differently. When we say data centers nowadays, what people are intuitively implying is there is a collection of similar boxes, hundreds, thousands, tens of thousands, maybe hundreds of thousands of them in a single data center. And the work is such that you can divide it up amongst all of these, and they can talk to each other if they need to collaborate.

**[0:26:13] Arvind Krishna:** That's the network or the optics that does all that. That's a data center. A mainframe, while it could be used in that context, really is. A mainframe is really useful when you have one piece of work that has an incredible volume. Example airline reservations. If we sell you the seat, you probably don't want to sell the same exact seat on the same flight to somebody else.

**[0:26:38] Bob Safian:** That would be inconvenient, yeah.

**[0:26:39] Arvind Krishna:** That would be so. That is a different kind of workload than you asking an AI model a question, and Joe asking it a question, and me asking it a question. That can be divided up because it doesn't need to know all the three answers, so the work is inherently can be divided or parallelized.

**[0:26:58] Bob Safian:** Part of the reason I I ask is because, you know, IBM was is sort of the the mainframe shop, the OG mainframe shop, right? And and there was a time where mainframes were sort of seemed like they were, I don't know, passe, everything was going to the cloud, and that has like shifted, like suddenly they're back. I'm sure the mainframe people don't like the idea of saying they're back, but but

**[0:27:19] Arvind Krishna:** So I remember in 1993, there was, I think it was Time magazine where they were showing a mainframe dressed up as a dinosaur, and it was called the death of the mainframe. It was only thirty four years ago, and then every ten years, people talk about the death of it. I think you got to be a bit more astute. What is the workload that is great for a mainframe? What is the workload that's not good for a mainframe?

**[0:27:42] Arvind Krishna:** I really am a believer in fit for purpose. The same way as a GPU is probably not ideal for running your smartphone because you kind of want your battery to last all day, not be over in three minutes. So there is a fit for purpose that is underneath these things. Where do you do AI training? That's one kind, where you do inferencing, that's a second kind, where you do web serving or streaming.

**[0:28:03] Bob Safian:** right, and well, and if you want to keep things secure, you want to have them on your own premises.

**[0:28:08] Arvind Krishna:** Sovereignty also comes into play because especially if you're outside the U.S. people care deeply about which government has control over the tech stack. So all of those things come into play for where you want to run things.

**[0:28:21] Bob Safian:** IBM recently announced 5 billion dollar initiative called Project Lightwell to identify and fix AI vulnerabilities in the open source world. And that was reportedly triggered by Anthropic's UM release of Mythos. What did what did you see that that sparked this at at that time? And and and what do you think people sort of misunderstand about cybersecurity overall?

**[0:28:43] Arvind Krishna:** So, first, for the good news, um, at least in our case, from the things that we've been running for the last few months, Mythos didn't find anything that other models didn't and couldn't find. I'll call that the good news. Here's the bad news. We have a lot of people, tens of thousands, who are experts in using these models to try to find vulnerabilities and then go fix them. Okay, so for the expert, they could already do all this using other models.

**[0:29:17] Arvind Krishna:** We will completely acknowledge that Mythos is way easier to use than the past. So what it did do was it opens up the attack surface to where I don't need one of those hundred experts to go do it. I can now do it with somebody with average skills. So that is definitely, uh, something to be worried about. So when Mythos came along, it's not just Mythos. The ability of these foundation models to actually help you write code, to understand code, is also there at the same time.

**[0:29:50] Arvind Krishna:** So the same thing which could be used to exploit, we could turn around and say, can I use it to fix at least all open source? And that answer became very quick, we can. So we said, as opposed to only worrying about, oh my God, I got all these things and like, okay, I have my list of 10,000 of them. We said, can we do something? By the way, it's not altruistic, it is good for society, but we do intend to charge, uh, people a fair price, not a not a usurious price for it.

**[0:30:24] Arvind Krishna:** To say, can we instead turn this into a utility where people can come to us? We can be a clearing house so that they can get their fix after giving us the vulnerability, but we can share th to others who are inside the closed set. Also that, hey, your friend here found a vulnerability, we're not gonna tell you which friend, and we're not gonna tell you where they're using it so that that information is anonymized and protected.

**[0:30:48] Arvind Krishna:** But you can actually get the same fix if you want. And yes, we are throwing a lot of people at it, but despite throwing that many people without using the current AI tools, it would have been impossible for us to go about saying that. If you give us a piece of open source, we can actually give you what is in our belief, a very well constructed, uh, patch, a fix against that vulnerability.

**[0:31:13] Bob Safian:** I mean, it's it seems like in this AI world, sort of s cybersecurity is like, my AI's gotta be better than your AI, right? Like the like, then the attackers are are using

**[0:31:24] Arvind Krishna:** It's the old, uh, I think it was Willie Sutton, right? Why do you rob banks? Well, that's where the money is. Why are you attacking cyber infrastructure? Well, today, that's where the data is, which is where the money is. That's why I began by saying the good news is that it's not really a brand new. The bad news is simply it'll be done faster. So nation states have been doing this for decades, but you would say three or four nation states were capable of doing it.

**[0:31:49] Arvind Krishna:** Maybe that opens up to a couple of dozen. That means more. So

**[0:31:52] Bob Safian:** and I guess it means organizations that might otherwise not have been targets, smaller, mid-size. It's

**[0:32:00] Arvind Krishna:** the potential function has come down, so it's easier to target more.

**[0:32:05] Bob Safian:** the bar has been raised, right? So we all have to be a little bit more prepared, even if, uh, even before we reach the scale of

**[0:32:11] Arvind Krishna:** I would turn around and say, if you don't think you're protecting yourself, it is only a matter of time.

**[0:32:18] Bob Safian:** it will come. Um, ear earlier this year, the the IBM Institute for Business Value released, uh, a provocative report called Enterprise in in twenty thirty and citing the big bets that CEOs needed to be making. And one of those bets was about quantum computing. Um, now we've talked here, most business leaders are struggling to adapt to AI. You're partnered with the U.S. government on a new quantum foundry, investing $10 billion in a large-scale commercial quantum computer.

**[0:32:49] Bob Safian:** W Why go all in on something even harder to understand and control than AI?

**[0:32:56] Arvind Krishna:** So let's go back to your very first question. If you can get ahead of the curve, and if what you're doing is hard enough that you actually have a couple of years advantage, our industry, the tech industry, has shown that you can create outsized returns for yourself and outsized returns for your clients in doing that. We felt that quantum is going to be one of those. We actually came to that recognition many years ago. Then the question became, can we do the hard science it takes to be able to make progress?

**[0:33:30] Arvind Krishna:** I would say earlier this year, we convinced ourselves of that. The evidence of that is both in our 10 billion dollar investment, because that means we expect to see a real return on it, as well as in the government agreeing to invest. Because that is a sign that they did their homework and agreed it is now time to scale this as an industry. So I think you should think of quantum as doing the following. CPUs, we've had them for sixty or seventy years, do a lot of great problems, right?

**[0:34:01] Arvind Krishna:** GPUs came around and did a different kind of problem. They did matrix math that allowed AI and other things to happen, but in some sense, it's not that CPUs couldn't do it, they were 10,000 times slower to do it. So last summer, summer of twenty five, they could simulate a five-atom molecule. I'll be honest, a five-atom molecule, a really good computational chemist, if it's a simple molecule, could probably solve by hand, maybe an expert, but they could do it by hand.

**[0:34:31] Arvind Krishna:** So you'd say, okay, your quantum can do it, but who cares? Last, uh, winter, November, December, they could do 300 atom molecule. That's good progress. Now you're getting beyond what you can do by hand, but you could do that on a normal supercomputer pretty easily. So you say, okay, you still haven't told me that this is interesting. In April, they did 12,000 atoms. They're now getting into the protein realm. When you're in the ten to thirty to 40,000 atom range, you're in the protein realm.

**[0:35:05] Arvind Krishna:** So this was a piece of a protein called trypsin. We are pretty sure that in another month or two, we'll be at double that range, which means you can solve trypsin. If you can understand the properties of a protein using a few minutes of computation, you can now understand which molecule, aka a drug, may bind to it to stop its bad behavior. We've now opened up a new pathway, possibly for health, which didn't even exist.

**[0:35:36] Bob Safian:** I mean, when you when you give that example, like part of the um amazing part of AI has been the exponential pace that it keeps improving. And as you give that example about quantum, y y, you're implying that it that is moving at that similar kind of pace. It's we're getting that much closer to not being science fiction,

**[0:35:56] Arvind Krishna:** Correct. So I think we are solving problems now, whether it's in. I talked about biology and molecules because I think most people intuitively get that that's a hard problem. But there are problems, things like fluid dynamics, aerodynamics, all of these are problems that are now coming right about now within the range of quantum computers to solve. Understand where quantum may be in two or three years, what kind of algorithms you may need to develop. So when it is there, you don't then spend two years doing all that.

**[0:36:31] Arvind Krishna:** That's what I would recommend to people to do today, but that makes it an easy. And that's not a a choice, then it's not a dilemma to say which of the two do you do.

**[0:36:43] Bob Safian:** and this, uh, th this, this, uh, report from the IBV, the enterprise in twenty thirty. Do you know what IBM will look like in twenty thirty?

**[0:36:51] Arvind Krishna:** We want to be known for not just being technologically innovative. I think that we have that reputation. We weren't always great at making that, uh, easily accessible to clients. So I think we want to be in the position where we are bringing all of our innovation to clients in a way that they can easily consume. That's one big piece. Two, we were about 20% software in 2019, we are now about 45%. I think that number will keep going up at a few percent a year, so we'll be much more in that space, uh, than anything else.

**[0:37:28] Arvind Krishna:** I think we'll become known as one of the exemplars of how we are deploying AI and agents to not just improve our own business, but to help improve our clients business.

**[0:37:39] Bob Safian:** I wanted to ask you one, uh, one last thing here. Um, a a a big focus for you is making IBM's culture more willing to take risks. For the business leaders who are here and and who are listening, w- are watching at home, like, w- what advice do you have on how you make that adjustment?

**[0:37:59] Arvind Krishna:** So I would tell everybody the most risky route is taking zero risk. What happens in any business that takes no risk? That means that you're kind of trying to extract profit, or what an economist would call rent, from what you already have. But that means you're giving everybody else the opportunity to clone you or copy you, be innovative from the bottom, so they will pick off the most profitable parts of your business. So now you have a declining profit pool.

**[0:38:35] Arvind Krishna:** Now you begin to have declining profit pools, and if you're conservative by nature, you're going to invest even less because you're getting smaller, so you're going to accelerate your decline, and you'll be approaching a cliff without realizing it. Go read history and see how many companies behave like that when you sit and watch them, right? It takes about five years and you begin to hit a decline, and then you're going along, and then five or ten years later, somebody comes along and either bites you up and chews you up and spits you out for parts, or you actually just go and fall over into oblivion.

**[0:39:12] Arvind Krishna:** So I call that that's the most risky. So how do you maintain enough innovation that you're actually growing? All innovation does not pay off. Innovation is risky by nature. So you got to say, how do I manage it where I'm generating enough profit to pay for innovation, recognizing that not all of it will have a long-term return, but enough of it should. They will pick off the most profitable parts of your business. So now you have a declining profit pool, and if you're conservative by nature, you're going to invest even less because you're getting smaller, so you're going to accelerate your decline, and you'll be approaching a cliff without realizing it.

**[0:39:54] Arvind Krishna:** Go read history and see how many companies behave like that. It takes about five years and you begin to hit a decline, and then five or ten years later, somebody comes along and either bites you up and chews you up and spits you out for parts, or you actually just go and fall over into oblivion. So I call that that's the most risky. So how do you maintain enough innovation that you're actually growing? All innovation does not pay off.

**[0:40:22] Arvind Krishna:** Innovation is risky by nature. So you got to say, how do I manage it where I'm generating enough profit to pay for innovation, recognizing that not all of it will have a long-term return, but enough of it should. Enough that's the nature of

**[0:40:39] Bob Safian:** I mean, people say, like, I, yeah, I'm fine with taking a risk as long as I don't lose anything.

**[0:40:43] Arvind Krishna:** well, this is humans are incredibly loss averse. If you're given the option of you can gain $10 but lose one, nobody wants to take that bet because they'd rather not lose one, even though you can gain ten. Okay, so we recognize that that's human nature. That means the point of leadership in at least corporations is how do you counteract that base human nature? Because people are not going to take risk if they think their jobs are on the line, or if you chastise them in public.

**[0:41:11] Arvind Krishna:** So what I always go to is I want your 50%, uh, probability win. I don't want the 90%, because if you tell people I need to be 90% certain, that's no risk. So you have to tell people, hey, it's okay that one of the two times you won't meet the deadline. So you start building a bit of a buffer in to say, okay, I got it that they won't meet the deadline. But if you're doing six things and four work out, fine.

**[0:41:33] Bob Safian:** And I guess you have to yourself own up to the things that maybe didn't work out the way you want, as you were talking about, you know, uh, Watson, maybe not working out the way IBM might have ideally wanted.

**[0:41:43] Arvind Krishna:** Plenty of things. I mean, I can think about our digital sales channel, something we're still working on. We're probably on our fourth try in my tenure, but we're not gonna give up it. It did work out means you gotta not just say, go harder at it. Not working out means think deeply about what is there structurally inside that didn't make it work, or is it that the market is different than what you presumed? You gotta sort of pressure test all those things and keep trying.

**[0:42:07] Bob Safian:** Well, um, Arvind, this has been great. Thank you so much for doing it.

**[0:42:11] Arvind Krishna:** It's my pleasure. Look, talking on these topics, I could probably go on all day. Thank you, Bob.

**[0:42:20] Bob Safian:** Thank you, thank you. I found Arvind unexpectedly candid from acknowledging IBM's strategic lapses with Watson to a looming AI data center bubble. And I appreciated that he could explain quantum computing's impact without getting too deep into the complex science of it all. What sticks with me most is his emphasis on risk, as he puts it, that the biggest risk is taking no risk. That's particularly true in times of tech transition, like right now. Thanks again to Arvind for joining us.

**[0:42:58] Bob Safian:** I'm Bob Safian, and thanks for listening. Masters of Scale is a wait what original. Our executive producer is Eve Tro. Senior supervising producer is Trisha Bobeta. Associate producer is Masha Makutonina. Video editor is Noah Wolstein. Senior talent executive is Stephanie Stern. Mixing and mastering by Aaron Bastinelli and Brian Pugh. Original music by Ryan Holliday. Our head of podcasts is Lital Malad. Special thanks to Jodine Dorsey, VP of Live Events. Visit Masters of Scale dot com to find the transcript for this episode and to subscribe to our newsletter.

**[0:43:49] Bob Safian:** And be sure to check out our YouTube channel.