IBM CEO Arvind Krishna says there is no AI bubble after all

Support for the show comes from Comcast Business. With advances in A I, cybersecurity is getting more powerful, but so are attacks. In the past twelve months, Comcast Business detected 24.5 billion phishing attempts designed to exploit user behavior, steal credentials and compromise devices. A I is forcing a shift in how organizations defend themselves, demanding not only faster responses, but continuous intelligence. To survive in the current climate, leaders must build a security first culture to equip their team with the tools they need to stay protected.

With Comcast Business, connectivity, intelligence and response work as one orchestrated system. Get better equipped to handle today's cyber threat landscape with the 2026 Comcast Business Cybersecurity Threat Report. Read it today at Comcast Business dot com slash cybersecurity. Cyber confidence starts with Comcast Business. Support for the show comes from Engine. Running a small business means every dollar has to work hard, but if your team is still booking travel the old way, it's costing you more than you think.

Engine is the fastest growing travel and spend platform in the country, built specifically for businesses like yours. Book a trip in as little as two and a half minutes, earn up to 10% back on hotels, and in 2025, Engine customers saved more than $300 million on travel with zero booking fees, no contracts and no B S. More than a thousand businesses join Engine every month. Join them and get $500 when your business signs up and starts traveling at Engine dot com slash Vox.

Support for the show comes from CrowdStrike. It's not a huge stretch to say that A I is the next major computing platform. Every platform shift changes the way we build software, and it changes security too. CrowdStrike is defining cybersecurity in the A I era with A I detection and response, or A I D R, a solution for seeing, monitoring and securing A I across the enterprise. Learn more about how leading companies are turning to CrowdStrike to secure A I and secure their business at CrowdStrike dot com slash decoder.

Hello and welcome to Decoder. I'm Nilay Patel, editor in chief of The Verge, and Decoder is my show about big ideas and other problems. Today I'm talking with Arvind Krishna, the C E O of I B M. I B M is a fascinating company. It's still a household name, and it's among the oldest tech firms in the United States. Without I B M, we simply would not have the modern era of computing. The company was instrumental to the development of a whole stack of foundational technologies in the twentieth century, and it still has a lot of patents to show for it.

But it's a lot harder for most of us to see what I B M has been up to in this century. Watson, the company's famous A I supercomputer, won Jeopardy in 2011, but since then, as far as most of us are concerned, it's been mostly ads during football games and not a lot else. I B M has been busy, though, just not in a way most of us can see. It's fully an enterprise company now, as Arvind explains, and that business is booming.

But there's huge change coming to the enterprise world as well. The A I technology that Watson pioneered, all that natural language processing and the beginning of what we now call deep learning. Well, that's given way to generative A I, and with it, a new way of thinking about how all the systems that run a company should be built and interact with each other. So I really wanted to ask Arvind how he felt about I B M investing in all of that Watson technology and showing it off a decade before everyone else, only to have maybe made the wrong technology bet and maybe now miss out on the modern A I boom.

I feel like Arvind would be pretty candid that the way I B M approached A I back then was off the mark. He says outright that pushing Watson so early into healthcare was quote inappropriate. But his take, as you'll hear him discuss, is that the infrastructure and research from that era weren't wasted because developers and companies can still build on top of that foundation. So, sure, Arvind says I B M got there a little too early, but he doesn't seem too concerned that I B M will be stuck on the sidelines.

Of course, I did have to ask him. The A I industry has all of the hallmarks of a bubble. Even OpenAI's Sam Altman is saying it's a bubble. Arvind's more optimistic than I am, or maybe less cynical. He's pretty confident it isn't a bubble. But you'll hear us compare the current moment to the dot-com boom and bust of the early 2000s, before the smartphone came along to realize the promise of ubiquitous online computing, and how ultimately disruptive all that was in some negative ways for a lot of people.

Even though all of the bets from the early dot-com era eventually proved to be correct. The other thing I wanted to ask Arvind about was if it isn't a bubble, who will win and what will that look like? Because it feels like Apple and Google managed to keep a huge amount of the profits from the transition to the online economy, thanks to their hugely successful ecosystems and app stores that effectively collect rent from the labor and transactions of almost every other player that has an app.

If the A I economy goes that way, will there be room for I B M or anyone else to get really big from? Arvind's answer seems to be that he's playing a different long-term game, which is where the company's big bet on quantum computing comes in. That bet still isn't making useful products for most people, but you'll hear Arvind explain why he has so much faith. This is a good one. He went all places and Arvind was remarkably candid.

Okay, Arvind Krishna, C E O of I B M, here we go. Arvind Krishna, you're the C E O of I B M. Welcome to Decoder. Nele, great to be here with you. I'm excited to talk to you. I B M is one of the most famous companies in the world. Candidly, I think most consumers don't know why anymore. It's it's very much an enterprise company. It have a lot of businesses. You have been there for thirty five years.

What has I B M been and what are you trying to make it today? You're right, I B M is an enterprise, it's a B two B company, to use a more common parlance as opposed to a B two C. Historically, I B M did do a lot of consumer products. We did the typewriter that people kind of knew it was iconic. Uh, we did the IBM PC, even though that's not been here for more than twenty years, and s a few other consumer things along the way.

But I would say, candidly, for the last thirty years, we've really had no consumer products. So what does I B M do? Our role is to help our clients deploy technology that makes their business better. And so as you think about that, whether they are on multiple public clouds, whether they want to take advantage of the data, they want to get to their customers faster, that's kind of what we're really about today. I think a lot of people know the Watson brand, which IBM has talked about for years.

Famously, Watson competed on Jeopardy. Now that I think the brand has turned into Watson X. There's there's a lot of what I would call like airport and football advertising around Watson, aimed, I think, directly at C I O s of companies and not at consumers. But we saw was still all experience at advertising. How does Watson fit into the I B M brand? Because I think that's what people really hook on to.

Yeah, if we think about the Watson brand, it did really well initially in putting A I on the map. The Watson uh computer won Jeopardy, and that shocked people. Oh my god, first time really that a computer could understand human language, think about sort of open-ended questions, and was more right than wrong. I don't say perfectly right, but more right than wrong. That woke people up, I think, to the uh possibilities of A I. And I will take credit.

I think it got us going on the current A I journey. Where it fell off was that we did things that were a little bit wrong for the market at the time. We were trying to be too monolithic, and we picked maybe one of the toughest areas, healthcare, to go into, which I think was inappropriate. The world is ready to take these things as building blocks. Engineers want to open it up, they want to see what's inside there, they want to build their own applications.

I want to use it for this, but not that. And so when LLMs came along, we got a chance to say, let's rebrand things, let's really rebuild the stack, and let's give people both the peace spots, but also a lot easier capability. And so that's what Watson X is. So it builds on the brand that Watson is associated with artificial intelligence, and I'm convinced that AI is a really big unlock for people. I gotta call it, it's the eighth technology, but that's a later conversation.

And so it's uh, that's what the Watson X brand is all about. So let me let me push on that a little bit. You described Watson as a computer, and it was a single computer that could go play Jeopardy, and then you described the introduction of LM technology and now this sort of ecosystem of building blocks where people can build stuff. What was the AI technology bet in the initial Watson computer, and do you think that that was the wrong bet as a technology?

Because I have a lot of questions about LLMs as a technology and a bet we're making, but I'm I'm curious now that you've had that experience. What was the technology in the initial Watson computer, and was it the right bet or the wrong bet? It's literally the same technologies that we talk about. So LLMs were not known at that time, but uh, sort of, I'll call it various other neural network models were. You know, neural network models all scam span all the way from what we would call machine learning to what was beginning to be called deep learning.

And so what was inside the uh Watson at that time was a mixture of machine learning, a lot of statistical learning, which became the core of what became deep learning. Let me just note the first big deep learning algorithm was a year after Watson won Jeopardy. Watson won Jeopardy in 2011, 2012 was sort of the the term came to be, but I would call it that the early incarnations of those things were in there.

Unfortunately, they were not there in a way that you could tune them, uh, take one out, make it modular and take another one. We were trying to give it to you monolith. That's what I meant by monolith. And that was the wrong approach, just to be straightforward, right technology, wrong go to market approach. Can you draw the connection between that set of technologies and LLMs today?

The the the counterargument that I would give to you is, you know, I'll just pick on Google. Uh, Google has had a number of bets across machine learning and deep research and LLMs for a long time. They shut off LLMs early, early. Like I I remember Sundar demoing, like, I can talk to Pluto and no one knew what he was talking about. And then three years later, ChatGPT happened, and Google was like, Wait, we invented all of that.

And that was their, you know, that was their technology bet, that was their paper. Attention is all you need. You're saying you had it, but I it feels to me like there was actually an inflection point where the industry picked a different technology, they picked LLMs. So can you just draw the connection for me?

For sure. So in the deep learning era, let's call that 2010 through 2022, that twelve years. Deep learning made incredible progress, no question about it. Here was the catch. Deep learning, to me, was incredibly bespoke. It could, you could take a lot of data, you apply a lot of people to label that data, it could do one task incredibly well, it really could. But tasks don't stay static. The data changes, the tasks change.

If I have to redo all that work of the human labeling, the relearning, the retraining, I'm calling that bespoke and fragile. So the return was always a little bit out there. If you have a massive singular B to C task, maybe suggesting which photograph you may love, or maybe r uh s picking which ad you may love, that applies. It's worth it because in a month or two months I can use that model, I can get a lot of return.

In an enterprise context, that's a little harder because it takes a lot more time to make up for all the costs that you put. So when LLMs came along, so to go back to the original work you referred to, uh, it was sort of intent and massive amounts of data, labeling goes away. Wow, that drops the cost by half. You do a brute force approach using lots more compute and lots less people. Wow, the cost comes even down because tech always gets cheaper over time.

So now half a dozen people, but using a ton of compute, could do what previously may have been thirty or forty phds and another forty or fifty engineers over six months. You can now do the task that much shorter. Okay, that's a huge unlock. So in the short, it looked like a two x advantage, a four x. But if I compare from the beginning to the end, this is a hundred x advantage in terms of speed and tuning and deployability.

That now looks industrial scale. So that means I can now take AI, plus these models can be tuned for many tasks, not just one, not all tasks. I'm not saying that, but many. That means that the applicability is massive. Also, when I want to ingest new data, I don't have to restart at the beginning. I can kind of add some. Yeah, at some stage, it makes sense to restart, but I can do a bit more there.

All of these are massive unlocks, which is why I think it's the right technology to help AI scale massively. By the way, I don't think it's the end all, but we'll come back to that. But it is a hundred times better than the prior.

that's the term that I'm really interested in, right? There were all these shots at AI before, deep research being one of them. There were machine learning algorithms deployed broadly across the industry. Apple was talking about the, you know, the neural accelerators in the iPhone years ago, but they didn't add up to what LLMs have added up to in the industry. I'm curious, though, you said cost, and you said cost can come down, but you know, we you and I are talking at the end of an earnings cycle, and everyone's costs are skyrocketing, right?

Their capex is skyrocketing. There's some layoffs associated with the increased capex that I do want to ask you about, but just purely on cost, it doesn't seem like it's that much cheaper, right? It seems like to win, you have to spend vastly more money, and that M money does not at the moment have a defined ROI. There's a there's a lot of bets. Can you reconcile the idea that it's it's less cost in this industrial scale versus the actual expenditures we're seeing?

Yes, I can, but I'm also going to, uh, if you'll allow me, uh uh, to say there's a difference in the B to C world and the B to B world also. So, first, let's just talk about the cost. Are there huge amounts of capital and and not just capital, but also operating expense being spent on populating data centers with GPUs, building out those infrastructures, and the amounts of capital being committed are now up in the trillions.

It's absolutely true, and that's what you just mentioned that, hey, that doesn't sound cheap, that doesn't sound a lot cheaper than before.

It doesn't even sound safe. I just to be clear, like, I don't even think that sounds safe based on the potential returns.

Maybe we'll come back to to that. What I meant by is going to get a lot cheaper. I mean, you if I take a five year arc, what has the semiconductor industry shown time over time over time? Go back to the beginning of the PC. You have half a dozen competing technologies, some begin to win. That was the beginning of Moore's law, really, right? Every two years you get a two x advantage in what you can do.

So I look at the semiconductor side, and I say over five years, we'll probably get a ten x advantage in the pure semiconductor capability, amount of compute for a dollar you can spend. Got it? That's one. Second, nobody has said that a GPU is the only architecture that is great for doing and deploying these large language models. There's certainly one. There are other companies coming up. We have a partnership with Groq, they have a different kind.

You have Cerebras, they have a different

so that's Groq, the the the processor company, not Groq, Elon's AI

Groq, the processor company. Uh, yes, the word comes from computer science. A lot of people use the word, but yes, sorry, yes, Groq, the uh, the inference, the inferencing chip company. And Groq is looks like, at least at our first uh steps, looks like it's ten x cheaper. But that, again, is not going to be the only design possible. So I think you get a ten x advantage on the pure silicon side, you're gonna get a ten x from the design side, and then the third piece.

I think there's a lot of work to be done around memory caching, how you deploy these models. Do I quantize them? Do I compress them? Do I always need the biggest? So there's a ten x from the software side. You put those three tens together, that's a thousand times cheaper. So I'm simply saying, hey, maybe you won't get all of it in the next five years, but even if you get the square root of that, that's thirty times cheaper for the same dollar to be spent.

If that, if that's kind of why I believe this is going to play out, it is going to get a lot cheaper, but it'll take five years to to play through.

five years, right right now, I think feels like forever to most people living through this disruption. It feels like forever when you can see the hundreds of billions of dollars being deployed today in data centers that are running, uh, mostly Nvidia GPUs. I look at all of that, and you know, you talked about Moore's law. I actually see a massive disincentive for Nvidia to come out with the next generation of its GPUs, right? There's a lot of equity tied up in the H one hundred being the U literal unit of currency that these deals are taking place upon.

Uh, that's a weird dynamic, right? It it, but it sounds like you say there's gonna be competitors that upend that dynamic.

For sure. It not necessarily upend, but provide a lot more competition, and that's the nature of it.

You kind of nodded in agreement when I said there was a disincentive for Nvidia to release the next generation of GPUs. Do you think that's true?

I think that always when you have an incredibly valuable company who's making its profit stream from a few products, there's an inherent or organic disincentive to try to modify that. That said, I would never bet against Jensen's ability to even disrupt himself and go towards the next uh plateau if there is one. So you have both. So I think certain companies are able to disrupt themselves, others hesitate to do it, and that is actually what causes the up and down of companies in the tech in the tech world.

I'm obviously leading towards the big question, which is this feels like a bubble. A lot of people think it's a bubble. You have a markedly different view of how this industry will play out. Uh, you're investing. I want to talk about the fact that you're hiring, while some of your competitors are are doing layoffs at at huge scale. But let me just ask the question directly, and then we can go into everything else. Do you think we're in an AI bubble right now?

No. Do I believe that there will be some displacement and some of the capital being spent, especially the debt capital, will not get its payback? Yes, but let's just sort of look at it. So this is a place, there is a B to C and there is a B to B world. There is a lot of common tech in both, but let's just look at the B to C. If in the B to C, you build a set of models that are very attractive and half a billion billion people become consumers of that, which is kind of the current numbers, roughly right?

If you build a slightly better model by spending another $50 billion, and that can attract another 200 million users, seems to make economic sense. And if you can then spend another fifty and get another 200 million into yours. So this is a race towards who can get more and more of the world's 7.5 billion people to become subscribers on a given model. Because then the next bet becomes, well, that network scale and those uh, e uh economies of scale will allow you to go succeed, and we've seen that movie play out.

That was social media in the last generation, going back twenty five years. So I react with it makes sense for them. Now, if ten of them are going to go compete, then we also know that maybe two or three of them will be the eventual winners, not all ten in that massive scale world. So to me, it makes economic sense that they're chasing that now. But my point being, not all of that will see a return.

By the way, if I go back to the fiber world, I would look at you and say, if I look at fiber optics in the ground back in the year 2,000, not all of those people got a return. However, this is the beauty of, uh, capitalism, and I'm calling it a beauty. We spend the money, it gets corrected back to thirty cents on the dollar. At that point, it makes an incredible amount of sense for somebody else to get that asset and turn it into a profit stream.

But it wasn't all got lost. As I said, two or three are gonna make a ton of money, and the others not, and so I think the equity being s uh being put in will actually get a return. Some of the debt will not.

I love the fiber comparison, and if you'll indulge me, I want to sit in it for just a minute. I was very young when the fiber rollouts were happening. I was very excited to get faster internet access, and I remember that bubble well. And part of that bubble was we're gonna build infrastructure for the internet, and the thing that really drove the bubble was we'll move the entire economy onto the internet. And that didn't work, right?

There was like a pets dot com IPO, and that was the sign that we hadn't quite moved the economy, but we built the infrastructure. And the important thing, and I think the important difference, is the fiber in the ground didn't go bad, right? We just had Gary Smith, who's the CEO of Ciena. They do fiber multiplexing. They can get infinite returns on that fiber that was deployed thirty, forty years ago to this day, and their technology helps them build data centers, right?

Which is really why he was on the show, because he really wanted to tell everyone that his technology could build data centers. The GPUs go bad, right? They're failing at a rate between three and nine percent already in the data data centers. And there might be an H 200 or the chip you're investing in with Groq might displace the H one hundred. So all of this capex is not gonna be here thirty years from now for the next generation of entrepreneurs like Gary to build upon and create more capacity with.

We're just gonna throw it away.

No No. So let's decompose it. So you're building a physical data center that's a lot larger. I think concrete and steel survive. Next to it is a power plant. We need the electricity. Actually, over time, I believe those power plants will even get hooked up to the grid, which is even better than for the national infrastructure. That's useful. Now, the fiber coming out of them, the networking inside these places, the storage inside these places, the CPUs inside these places are all useful.

I'll acknowledge right now there is a very high failure rate. But being a bit of a semiconductor geek, though, I'm not anywhere near as deep as some of my friends and competitors in those spaces. If you can run something at three gigahertz, if you try to run it at four, it will actually run, has a higher failure rate. If you try to run it at 300 watts of power, maybe it's great. If you run it at 400, it has a higher failure rate.

So if today you just need the performance so you can train a model that much faster, it actually is worth it to tune it. That I'm okay to have that failure rate. I got software that worries about moving stuff around, but you can detune it slightly and get to higher resilience. So I think that is actually a design point. That's not really a bug, so to speak. Now, over time, do I acknowledge that these will move up?

I began by saying, I think in five years, our semiconductors will be a hundred times better. So you're right, there's a five year depreciation to the GPU or the sort of the compute infrastructure, but the other half is useful. But in five years, you don't throw away all the capex, you throw away a little piece and the you replace that with something that is better at that point.

I think the the specific comparison to fiber making, and maybe it's too pedantic, but the fiber was in the ground and then it was there, right? It did not incur a recurring cost to the people who wanted to use it. Outside of, can you create more capacity by multiplexing the fiber?

No, P. Y- you're right. The fiber in the ground is endurable. Maybe not forever, but at least for a hundred years. At some point, even glass actually begins to occlude and do all kinds of weird things, but it's good for a hundred years. But people also built a lot of end stuff on top, all of which had to be thrown away. I mean, like, you're now forgetting all the failures. People are building ATM. People thought that they would have built really intelligent, uh, video streaming and put that, uh, the guts of that inside.

People were talking about doing wavelength division multiplexing since you talked about Ciena, and the end became simpler. Here's dark fiber. It's a dumb pipe. Go throw your bits in it at a terabit. The intelligence belongs at the cloud end. That took ten years to unfold. So there was actually a change in how it transpired. I mean, I'm sorry to be that geeky about it, but

That's fair enough. No, I'm here. This is why we're here. This is why I ask the questions. We have to take a quick break. We'll be right back.

Support for the show comes from Vanta. As AI adoption grows, so do your company's security risks and requirements. Most compliance tools promise automation, but you're really stuck doing all the work by hand, tracing screenshots, validating controls, and playing catch up when audit season comes around again. Vanta works differently. Their agentic trust platform is built to scale with you, not slow you down. With over 1,400 automated tests across more than 400 integrations, Vanta collects evidence and monitors your controls year round.

Plus, you can access the Vanta agent everywhere you work, whether your team is in Vanta every day or living in tools like Claude or Cursor. They work around the clock from full program context, so nothing slips through the cracks. The numbers back it up. Vanta delivers 526% ROI over three years and pays for itself in three months. Over 16,000 fast growing companies like Ramp, Rider, and Harvey save time and resources while scaling fast with Vanta.

Learn more at Vanta dot com slash decoder, spelled V A N T A dot com slash decoder. Support for the show comes from Shopify. Starting a business requires you to be resourceful, and the more you flex that muscle, the more you start to see opportunities for growth. For example, launching your business with Black Friday approaching means you can take advantage of the extra online traffic. Shopify is there to help you make it all happen. Shopify is the platform for a record breaking, reliable holiday selling season for businesses at every stage.

Their pre built website themes and AI site builder make it easy to launch a beautiful, professional looking store. And you don't ever need to touch a line of code to do it. They also have a mobile app that lets you run your store from anywhere. Their AI e commerce expert Sidekick can help you manage almost any business task with a single prompt, including holiday discounts. With Shopify, you can easily sell everywhere people shop online, in person, across social media, and even in AI chats.

If you're ready to launch your business before Black Friday, go to Shopify dot com slash decoder to start your free trial. That's Shopify dot com slash decoder. Go to Shopify dot com slash decoder today to hear your first. Support for the show comes from Granola. Have you ever left a meeting thinking that was productive? And then ten minutes later, when it's time to take next steps, you're at a loss. Or worse, you've been heads down the entire meeting, furiously typing notes instead of actually listening, and now your brain is fried.

Meetings are a mess. Granola fixes that. Granola is an AI powered notepad that makes your meeting notes useful. It capture what happens in your meetings and turns it into clean, structured notes with decisions and action items pulled out and made easy to find. And the best part, Granola works through your device's audio, which means it integrates seamlessly into the video conferencing tools you already use. No setup, no awkward bots. It's just your normal meeting with superpowers.

You get to actually listen instead of frantically typing every word, and still walk away knowing exactly what was decided, who's doing what, and what comes next. Once you try it on your first meeting, it's hard to go without. You could try it totally free. Just head to granola dot ai slash decoder. That's granola dot ai slash decoder to get your time back. Go to granola dot ai slash decoder and try it on your next call. Support for the show comes from Engine.

Running a small business means every dollar has to work hard, but if your team is still booking travel the old way, it's costing you more than you think. Engine is the fastest growing travel and spend platform in the country, built specifically for businesses like yours. Book a trip in as little as two and a half minutes, earn up to 10% back on hotels, and in 2025, Engine customers saved more than $300 million on travel with zero booking fees, no contracts and no B S.

More than a thousand businesses join Engine every month. Join them and get $500 when your business signs up and starts traveling at Engine dot com slash Vox. Welcome back. I'm talking with IBM CEO Arvind Krishna about the infrastructure of it all, and whether all the stuff IBM is building is really as bubble proof as he's betting. I would actually argue that was one of the most exciting periods in tech when no one knew how it would work, and there were many, many more shots being taken.

It all did pop in a tech kind of shot in a bubble, but it was very exciting. It did go down. It did go down, and then today you turn around and say, but all the companies that got built on the back of that clearly prove that that investment was worthwhile. If I look at it at a national or an aggregate investor level, while some people did lose a lot of money, by the way, some people made a lot of money.

I want to take the other part of that bubble comparison, which is we're going to move the entire economy to the internet. You brought up social media. I would broadly characterize that as someone who covered it very, very deeply from the beginning of the iPhone to now, is we're going to move the entire economy onto your phone, right? There was first, we're going to put it all online, and maybe we don't have the distribution because we're not all going to look at CRT monitors on our desktop, and that didn't happen.

But then we all got phones, and the idea that we could move an enormous amount of the economy, the consumer economy at least onto our phones, it happened, right? I mean, it that occurred, we, we're all living with the results of that today. Do you feel like the argument, at least in the consumer space, as you've described it, is we're going to move that app economy to AI? Because that's how I see it is a bunch of people who got rich moving the economy onto smartphones.

It's that same class of investors who now think they can run the playbook again with AI, and maybe we'll re rearchitect the applications with MCP, and maybe there'll be agents using the websites instead of people.

But the argument from the same set of characters feels broadly the same to me. If you don't mind, maybe I'll go a little bit deeper onto your first part.

You're absolutely correct that the front end of the economy kind of moved onto the phone. The moment the phone gave you access, uh, so that you could, it would could be with you everywhere. You were not just anchored to a desk with a laptop or a desktop. It definitely was a massive unlock. Let's acknowledge that, right? But there is a physical economy still. I mean, I always talk about 60% of the workers, even in the United States, are still frontline.

People who do construction, people who have warehouses. If you're buying a tangible good, it's still coming from a warehouse, maybe not from a retail store near you because they had a front end, but in the back there's a warehouse and a truck driver and maybe multiple routes of distribution. Uh, we still go to restaurants, there's still a food, there's still groceries, there's physical healthcare, there's all of that. It becomes more efficient, easier, more convenient. So if I now say, I don't have to spend that much time, I'm gonna have an agent or an AI front end that helps unlock even more and puts together four or five things that I have to put together in my head.

I completely agree with you. Why wouldn't we want that to go happen? That is going to go happen. You can see the early instances of that already happening. And now why it's so appealing is because it gives a chance to people without me taking any names to reform who are the biggest players. And it gives a chance that there will be some disruption. And then the other side, I actually think it goes beyond consumer into the enterprise.

I actually believe there's going to be a billion new applications written. Now, if you think about the the smartphone ecosystem you talked about, people talked about half a million, a few million. I think this could be a billion, because there may be a few million that sit on the consumer side. And then if every enterprise has, let's call it a thousand, you go across the number of enterprises times a thousand, then that unlocks a lot more, if that makes sense.

it does. Let me ask you one question there, and then I do want to ask you the decoder questions, and I want to talk about IBM specifically a little bit. Uh, that move, we're going to put all the economic activity onto the smartphone. In many ways, th that economy of applications, the biggest winner were Apple and Google, because they collected an enormous amount of rent on the back of that transition, right? They the app store taxes and the fees, and maybe that's going to get unwound now with whatever antitrust litigation or whatever is happening in Europe.

But it happened, right? They collected a huge amount of fees. They are some of the richest companies in the world on the back of that. Apple just reported its, uh, quarterly earnings that their services revenue is higher than ever on the back of app store fees. Like that, that's what that line really is. It's it's not, I think they run the TV business just to pretend that that's the reality is not the reality. Do you see that playing out in AI?

Because I look at OpenAI announcing what looks like an app store. I look at Google announcing the Google search will have, you know, inbuilt custom developed applications as you search. That it's it's very cool, but I see, oh, we have these points of centralization emerging again that don't look like Apple and Google, and maybe there's competition for that. There might be competition for that in the enterprise. But do you see those point, this same, those same points of centralization?

I wouldn't jump to that. We know who's the winners today because we're kind of in the first innings of the game. There will be some winners. How about if I agree with you on that?

But do you think those winners look like those central points of control that we saw in the smartphone era?

There will be a few different. So if you go back to the smartphone analogy, you had one who built a vertically integrated stack. It was a easier, more convenient device. And then to get access to that device, people had to come into the app store. And so that was that model. The other model said, we are completely open. It's an Android, um, was the operating system. However, to get access to everything else, now you come into the app store or you come into the search.

And that was the second. It wasn't an identical, but it was a a similar. And so that those became the two entry points, so to speak, to get access to the end individual. And that's why they could charge their appropriate, you're calling it rent, which is from an economics term. Let's say they could charge an appropriate margin, maybe from a business.

sure, yeah, I think Tim Cook would call it a margin, but I the developers I know feel very differently about that margin. The reason I'm asking,

I'm not, but but those who build out that massive infrastructure, there is also a massive amount of cost that goes into. It's not like they can maintain it forever. I mean, as the Chinese have shown, you can build competing products. If you can keep running ahead, then people will prefer these devices. But if at the end of the day, the value is in the apps today, as you are saying, then if that app is available on something else, if the friction and if the innovation in the main platform slows down, people will switch.

It'll take maybe three to five years. So it's not like it's a guaranteed return forever. Uh, it will switch. And as many other companies have seen, that switch takes a few years, it doesn't take decades. When it happens, though, is disastrous to the to the original company. Some manage to recover because they wake up and say, Hey, wait a moment, I got to change. Some don't.

I think this brings me to to IBM, right? This is the process you're in with IBM, and IBM has been in for many years now. Uh, you took over as CEO in 2020, you'd been at the company for almost thirty years when that happened. Uh, I ask everybody these questions. You, you have a unique perspective here. You'd been at the company for a very long time. You took over as CEO. How was IBM structured when you took over, and how have you changed that structure now?

It's much more about culture, focus, and what we do and how we do it than sort of the formal organization structure. So if you say that you got to be focused on innovation, and you got to be focused then on where you can provide a unique value back to your clients. So that's kind of the first question. And I am clear our sweet spot is helping our B2B clients succeed. So you say, okay, well, that's a very big remit.

What in there? So I held two points of view that were somewhat unique. One, I don't believe that the majority of our customers are going to go to a singular public cloud. Some will, but the majority will not. First of all, people outside the US tend to want to be somewhat, uh, split between maybe an American cloud and something more sovereign. Okay, then there are people who use plenty of SaaS properties. There's a huge amount of economic value in what they've already written in their pre existing applications.

I'll use the word hybrid to describe that. Is there a place for a vendor to have absolutely leading edge tech to help our clients in that journey? So that's the hybrid approach that we took, and that clearly over time has shown to be of incredible value. About 60% of the total spend is outside the US. Even inside the US, anyone in a regulated industry is going to be hybrid in some sense. So that's kind of the first.

The second was, let's focus on where AI can be deployed in the enterprise. Let's not go try to go compete. I will not try to compete with Google on building a chat bot that goes, What's the current number? I think they talked about yesterday, 650 million active subscribers. That's not where we have brand permission and brand credibility. But can I walk in to a health insurance company and say, I'll make sure that your your clients, in this case, patients, health data is protected, but let's help AI unlock to make those people feel even happier, get quicker answers, easier answers.

And we tend to have the trust with those people that we have never, in 114 years, misused that data, not even once. So you get that, and then you can go give them the tech and get that deployed. So we picked those two, and then I said, what are we really good at? We're really good at building systems. And so the third bet, uh, that I decided early on is we're gonna make a bet on quantum.

Now, let's see whether we can get it from being a science to being an engineering challenge. And then once it's an engineering challenge, then how do we scale it to really get deployed? That was really the big inflection we made. So as opposed to trying to do lots of things. So that meant I use the word innovation. So that meant commodity services have to leave the company because you can't do both. It meant if we are going to be hybrid, I got to partner with everybody else that I talked about.

So you begin with the clear view of what should be done, and then you say, then it doesn't matter. I'll make all the hard decisions of changing the way the sales teams are paid, of changing the incentives of all the executives to align with what's then needed to make those things succeed. Sorry for a really long answer.

Yeah, no, it's great. I uh, what one of the things that I think about all the time, what a trope on this show is that if you tell me your company structure, I can predict like 80% of your problems, right? You might say culture and structure are divorced, but I see the connection and they they feed off each other. So you were at IBM for a long time. Vanishingly few people will ever interview to be the CEO of IBM.

What was that process like? Did you come in saying this company is focused all wrong? We got to let go of the commodity stuff. I'm gonna make these changes. And then once you had decided to do that, how did you actually change the structure of the company to focus on those things?

I probably didn't spend thirty years aspiring to this job, just to be upfront. I think it was more of a process of discovery, even for myself in the last couple of years before that. So I made the hybrid observation deeply in 2017. As I was making that, I said, Okay, how do I test this? So I actually had a partnership with Red Hat, and I said,

Is this why you have a red hat? I noticed you have the red hat behind you.

I have a red hat there because remember when we made that decision, that was 2018 that we announced. It took a year to get through regulators to close it. It was 30% of our market cap. Very few companies spend 30% of their market cap on a conviction and a belief. So I keep the red hat there because to me it was clear, if that conviction turned out to be wrong, I should be fired.

I actually don't. People hesitate to say those things, and I say, if I'm that wrong, I should not be working here. And that is why I keep the red hat as a reminder to myself that not only must you have the conviction, then you must do the really hard actions that take. So that's the culture part of making that conviction succeed, because otherwise people will just fall back into the lanes they were in. There's comfort in doing things, um, the way they've always done them.

But but just put me in the room. It's it's 2020, you're going through the interview process with the board. Did you have a deck that says we're doing too much commodity stuff? I, I, I'm gonna cut it down and we're we're gonna focus on these areas and the big bet is that change that quantum would be.

My deck was like my deck was three pages of prose. It was not like a hundred pages of analysis. I actually believe that you should talk about what you want. I said, we have to grow, and my view is very simple. You got to grow well above GDP growth, otherwise you're not going to be that relevant into the future. Okay, if you're going to grow, where are you going to grow? We are. If you look at us, our brand permission is fundamentally being a technology company.

Okay, that was code for high innovation. So that means now this is where people, I think, uh, f, I think many companies fall short. If you're clear about that, then things that don't belong in that should not be in the company. So that is why the spin outs that we did, which did take a couple of years to get done. Then I said, we have to grow in software, because that is where value is perceived by our clients.

So if you're going to grow in software, so you talk about structure, but that becomes a big fundamental change. That's where capital allocation, that's where resource allocation goes, and that's where you got to put way more investment than you historically have. Then if you're going to go with partners, how do you fundamentally line up with partners? So that is organizational change because you got to say, how do the sales teams get paid? How do you have the right incentives?

So those are sort of maybe the three first really big decisions, uh, for me in the first two years. As you do that, you also realize people tend to be very risk averse. How do you unlock them into taking risk? To me, there's no risk free path to success. If you want to be risk free, you're going to be slammed against the bottom end of performance almost always. So how do you unlock risk taking in people in a way that they feel motivated to do it more often than not?

This leads me into the second decoder question. I ask everybody. I have a sense of it, but I'm curious how you will describe it. How do you make decisions? What's your framework for making decisions?

You've always start with is there a value? If it's a decision that's going to impact what we do and how we do it, does a client benefit from this new way of doing it? If you're pretty convinced of that, and I'll come back to how you kind of get your conviction around it. I always believe that one should triangulate. So I always will talk to a number of people on the inside, on the outside, not maybe with a full description, because sometimes you don't want to give, but enough to validate my assumptions or my what could the possible victory be if we made that big decision?

You arrive at a conviction, you triangulate it with a few people, and then you ask yourself the question, what needs to change inside if we really want this to go all the way? And once you arrive at conviction and all those, you then are able to go execute. I build on my own strengths, so I think I'm a reasonably deep technologist. I think I generally can understand where the tech can go, but I may not always fully understand what a client can do with the tech.

That's why the first piece is really important. Then I triangulate because, hey, I don't know, like, talk to people, talk to somebody inside. I don't mind at all reaching like ten levels down in the org to talk to somebody who I think, uh, has an opinion on that topic or knows about it. Talk to possible clients about it, talk to partners about those things. It just informs your opinion, and maybe even more than that, when you're out talking to them in any case, keep your ears open for what they say that could actually inform some of those things later.

We have to take another short break here. We'll be back in just a minute.

Support for the show comes from Framer. The future of website building is seamlessly blending all the tools we have available, and you could use a platform that has it all and stays useful even after you launch. That's what Framer is. Framer is the AI website builder that helps creators, teams and businesses ship production ready sites faster than ever while getting every detail right. You're able to prompt, inspect, edit, and publish in one place at a whole new pace.

On Framer, agents bring speed and scale, while people bring taste, judgment and control. And that all takes place in the same workspace where the real site is designed, managed, reviewed, and published. Everyone's ideas land on the canvas, stay editable, and can be published when the team is ready. Thousands of businesses, from early stage startups to Fortune five hundreds, are choosing to build their websites in Framer, where changes take minutes instead of days. Learn how you can get more out of your site from a Framer specialist or get started building for free today at framer dot com slash decoder for 30% off a Framer Pro annual plan.

That's framer dot com slash decoder. Rules and restrictions may apply. Support for the show comes from Framer. If you want to create a website that looks great, feels handcrafted and comes together fast and seamlessly, then look no further than Framer. It's a complete website platform, not just a builder, so teams can launch and keep improving their sites in one place. It helps creators, teams and businesses ship production ready sites faster than ever while getting every detail right.

You're able to prompt, inspect, edit, and publish in one place at a whole new pace. On Framer, agents and humans work in tandem. Agents bring speed and scale, while people bring taste, judgment, and control. And that all takes place in the same workspace where the real site is designed, managed, reviewed, and published. Thousands of businesses, from early stage startups to Fortune five hundreds, are choosing to build their websites in Framer, where changes take minutes instead of days.

Learn how you can get more out of your site from a Framer specialist or get started building for free today at framer dot com slash decoder for 30% off a Framer Pro annual plan. That's framer dot com slash decoder. Rules and restrictions may apply. This holiday, switch to the Simplicity Plan from Verizon. It's one plan, one price, just $30 a line per month for you or the whole family, and on their best five G network.

Switching to Verizon and shopping their best deals makes the holidays simple. Head to verizon dot com or your nearest Verizon store today. Black Friday is around the corner, your biggest chance to grow sales this year. Don't wait. With Realize, Taboola's advertising platform for businesses, you can reach people engaging with your competitors across premium websites with 600 million daily active users. And as an exclusive offer for podcast listeners, new customers get $700 credit back.

Just search Realize Podcast on Google or use the promo code Podcast twenty six. Ready to get started? Take advantage of this offer while it lasts. Terms apply. Welcome back. I'm talking with IBM CEO Arvind Krishna about IBM's biggest, weirdest, and maybe most interesting bet of all, quantum computing. Let's put that into practice on kind of the farthest bet you're making, which is quantum. All the big tech companies have quantum divisions. I've had Jerry Chow, who runs part of your quantum team on the show before.

That was a great conversation. I've looked at a lot of rooms where someone tells me that this is the coldest place on earth to run their quantum, whatever qubit they're trying to generate on that day. None of that has paid off yet, right? We're not at at close to what they call utility scale computing in quantum. That's not something your customers are asking for yet, right? Like, that's outside of the purview of the structure and the culture you're deciding.

That's a big bet that there will be a massive step change in how we build computers that unlocks vastly more value for everybody. And you have to keep that investment, even through all the turmoil and all the data center investment that everyone else is doing. And, you know, Amazon saying we're laying off 14,000 people because of AI, while you're saying we're going to hire more college graduates than anybody else. W What is the decision to stay focused on quantum in that way?

How do you how do you maintain that decision?

You are right that you can't go check it with a customer because they don't know what to do with it today. But that's not fully true. So the first five years, 2015 through 2020, you got to have a belief of what it could do.

And maybe because of some of my graduate school math background, it was like, Wow, if we can do that, I can immediately see what kind of problems could get unlocked. But trying to explain that to anybody but the people excited in the field is impossible. So let me completely acknowledge those five years is about a internal bet on a set of people and a possibility. Okay, but 2020 onwards, we began to say, got it, these are not utility scale, let me acknowledge it.

They're full of errors. They are small. Clients still get excited by it. So I did do a full check. We have 300 clients, not commercial. There are 300 people working with us in a, let's call it, research mode. A hundred that are pure, uh, in the end, commercial, a hundred who are in the world of materials or medicine, and a hundred who are pure academics, right? Rough pockets. That's why HSBC proved to itself we could do bond trading pricing on it.

Vanguard proved to itself that if they get big enough, we could do a portfolio that better appeals to your needs. Uh, you have, um, Daimler working on EV batteries. Um, I have, uh, Boeing looking at corrosion on materials. So there, so there is a proof point. They're not saying they'll buy it the way it is today. All they're saying is, hey, if you get to that point, this is really interesting to us. So there is a validation, even from clients.

And then I said, How do I know there's enough interest? So I asked the team, put the software out open source. Now, I'll say for many people, including some of the current AI, that's not a common this thing that you do early on. I said, why open source? How will developers and universities use this stuff and get any excitement if you put a price to it? So we put out all our software open source. The fact that there's 650,000 people globally who use it, hey, it tells me that there is excitement and there is a movement, and people are hungry for a new approach to do other kinds of problems.

So those were the two validations on my framework that were useful. If that 650,000 had been 100,000, I might still be okay. The fact that it's 650 tells me there is real, real traction. But if 650,000 had been one thousand, I would have told my people, right? Guys, this is your physics friends. This is not a market.

So I I'm curious about that, right? That that is the kind of long term bet and a sort of early interest from people who think, okay, this type of computing will let us do many more things. And you know, it's funny on the consumer side, I hear about it in terms of, well, when there's quantum computing, we'll need quantum proof encryption. And it's like there's a secondary market now, based on whether or not you will succeed in quantum computing.

That almost has nothing to do with the quantum computing succeeding. It's it's. It's a bet, right? It's like a. It's it's a strange hedge against your success or Microsoft's success, or whoever else is doing quantum. What does the actual success look like? Is it a step change in computing that is as big as we're gonna re-architect all the computers around AI that we're experiencing today? Is it bigger than that? What does that feel like to you?

I actually think that the it's an ad, so if I call it the CPUs, GPUs did not replace CPUs, it was an ad. Now, GPUs are priced much higher than CPUs, so the market is bigger for GPUs than CPUs, but it was a complete ad. It didn't displace what AMD and Intel and Arm do, just to be straightforward.

I feel like Intel feels differently about that right now. Sure, Nope, I I agree with you. means have many Yeah,

other issues than the number of x86 chips being sold per year is as high as it has ever been. How about if I phrase it that way? fair enough, yeah, Okay, so it's an ad, but the next one, if it has more immediate value, you can price at a different price point. Does that make sense? I think G is quantum, so let's just use the word QPQPUS, just to keep it simple. When they come, are going to have an incredible value, because they can solve problems that you actually cannot solve on GPUs and CPUs in any economic terms in the near term.

The same way, look, everything you can do on a GPU, you could do on a CPU, just to be straightforward, but it's going to be a thousand times slower and not be maybe economically feasible. So GPUs opened up a whole class of new problems. QPUs are simply going to open up. It's an AND. It's not a displacement. But given there's finite dollars in the world, if there's an AND and we have a first mover advantage, like one of the companies you named did in GPUs, that opens up a possibility that the market is that big.

So we did work because I don't want to my point of triangulation. So we asked a couple of our friends in the consulting world, BCG, McKinsey, hey, tell us, if we can arrive at some utility point, what do you think the value is? They both came back and gave us a pretty consistent answer, very slightly, but think of it being we think there's four, five, six, 700 billion of value in the early years per year.

Great, how much do you think the tech world could get out of that? Probably 20% to 30% seems reasonable always. I said, OK, that's the size of the prize that we're going to chase. Now, how much of that share will we get versus others? It's sort of the question, and that's the journey we are on for the next five years.

so five years, you think you will be able to pay off the quantum investment in five years?

It's really hard on hard engineering to put like a dart and say, you know, this is not like building the next mainframe. There, I really know what I'm doing. I know exactly how many months it'll take and I can put here. I kind of give it a spectrum. Will we get to something remarkable in three, maybe three and a half years? I'm gonna give it low odds. It's possible, but it's maybe twenty, 30%. Can we get there in four years?

My odds go way up. Can I get there in five years? My odds go really high. So that's why I say five, not not to say it's really is there. I think it'll be a bit of a spectrum. You'll see some really early adopters come in, I'm hoping in three to four, then there'll be more at the end of four, maybe, and then the risk decreases for people after that.

that's a lot of action in twenty four months. I'm I think that that'll be a very exciting two year period if you hit it. This is really interesting talk about in comparison to AI, right? You're talking about how you estimated the market size for a nascent technology that you have to develop, like actual capabilities for. You estimated how much of that market share you could take. And then you, you're obviously making some investments based on the potential return.

And the last part, why us? If I assume all of this, which I talked about, others could do so, then why would we succeed? Because I think it's much more. There's so much talk. And you mentioned the various qubit technologies and cold rooms and alternate technologies, and I actually love the fact that there is that much. But that's not building a computer. And I always tell people, you need a great QPU and a great qubit. Absolutely, you also need a way for them all to talk to each other.

You also need a way to go control all of them. You also need a way that it functions by itself, without six quantum physicists standing in the room doing it.

Now, this is a great employment plan for quantum physicists.

So you need all those things, and I think we are one of the unique players who have a lot of those skills in-house. So it unlocks people to go do that and really, really motivates them and excites them. And I think that that is, that is an advantage. I think we have today in terms of the underlying skills.

I would call that a very sober, very thoughtful, conservative almost approach to deploying billions of dollars in capex against a technology that has not yet proven itself in the market, right? You've you've made some estimates, you have an idea of what your company can do to add value. You're gonna do the hard research and then you're gonna get there. I would just compare that to OpenAI, right? In the AI market that we see today, just this week, you know, OpenAI, they converted to a for-profit.

There's reportedly 1 trillion dollar IPO coming. There's everything we've talked about in the enterprise space, where you can see maybe how AI and enterprise can help accelerate use of data and all this unstructured data that companies have. Fine, but the bet is in the consumer space, we're just gonna build a full-fledged agent that's gonna run around and do stuff for you, and that will replace your smartphone. And none of that seems sober or conservative, or based on a a real market estimate, or even whether consumers want that product, right?

It's it's just a pipe dream. How do you reconcile those two things? Like, really, the bet is there will be a G-I. Like at the end of the day, that is the the whole market is. Someone's gonna figure out AGI, and then all this will have been worth it. The the press release from Microsoft announcing the restructured deal with OpenAI several times mentions in bullets that the terms expire when OpenAI declares AGI. I read that, and I thought that this is the most remarkable press release I've ever read in my entire life.

No, no one can even define this term. And now two of the richest companies in the world are are issuing press releases saying their deal will restructure itself when that happens. That's very different than your bet on quantum. How do you read that discrepancy?

Of the ones you mentioned, one has a huge amount of cash flow and ability to invest. So for them, it's not existential,

That's Microsoft.

like something that could be incredibly profitable. The other one is is a classic Silicon Valley startup. I'll put it that way. Some will succeed, some will not. I'll offer you an opinion. First, I don't think deeply about the whole consumer side and how much money they'll spend. It's interesting to observe, but I'm not going to pretend that I deeply.

Do you? Well, let me ask you this question Do you think there's an enterprise ROI that would justify the spend we have today? Cause I look at it and I say, absent AGI, this spend might not be worth it.

You said I'm a little numerical, I'm a little geeky, I'm a little like, I.

I'm, I'm having the time of my life. Navigate Just like a vision, by the way, I love it.

so at today's costs, so let's just ground in that, because anything in the future is speculative. It takes about $80 billion to full to fill up a one gigawatt data center. Okay, that's today's number. So if you are gonna commit twenty to thirty gigawatts, that's one company, that's 1.5 trillion of capex. And to the point we just made, you gotta use it all in five years, because at that point you got to throw it away and refill it, right?

Then, if I look at the total, this thing's the total commits in the world on this space of the chasing AGI seem to you like a hundred gigawatts. At these announcements, that's 8 trillion of capex. There's no way you're gonna go get a return on that is my view, because 8 trillion of capex means you need roughly 800 billion of profit just to pay for the interest.

Have you told Sam? Cause he I, he seems to think he can get both the

I got it. capex and the But that's a belief. return. It's a belief that one company is going to be the only company that gets the entire market. I got it, that's a belief. That's what some people like to chase, and I understand that. And from their perspective, that's different than agree with, understand is different than agree. And so I think it's fine. I mean, like, they're chasing it. Some people will make money, some people will lose money, and all the infra being built will be useful if it goes away.

But it's if they make it, then they are the sole surviving company. I am not convinced, or rather, I give it really low odds, like we're talking zero to 1%, that the current set of known technologies gets us to AGI. That's my bigger gap. I think that this current set is great. I think it's incredibly useful for the enterprise. I think it's going to unlock trillions of dollars of productivity in the enterprise. Just to be absolutely clear, that said, I think AGI will require more technologies than the current LLM path.

I think it'll require fusing knowledge with LLMs. And we have words. I'm not sure that's the only way to create knowledge. People talk about neurosymbolic AI, but I think if I just say knowledge in a broader sense, hard knowledge that people have spent thousands of years discovering. If we can figure out a way to fuse knowledge with LLMs, maybe even then I'm a maybe I'm not like 100%, but that's that's just sort of, uh, sort of a a geeky

no, I'm.

technical view.

Actually, that was my question, and you started answering before I asked it. I'm on the same path as you. I think I look at what LLMs can do today, I look at how people talk about the scaling laws, they might hit, the need for more data that doesn't necessarily exist at the scale it might be needed. And I say, I don't think LLMs can do it. Like, I don't see a, you know, here to their path for this technology to get to what everybody says it can do.

And I would just connect that to what we started with, which IBM developed Watson, and it was very good at its tasks, and it wasn't the right set of bets at that moment and you had to pivot it. Do you see a next technology that LLMs would have to pivot to, or that the AI industry would have to pivot to?

So, so let's look at three, maybe. Machine learning was not actually replaced. Machine learning is incredibly useful for lots of simple tasks. Your little thermostat thing in your house uses machine learning, not LLMs. People look at like golf ball trajectories, baseball, uh, tennis balls, that's all machine learning. It's not being replaced. Really useful, but it's not gonna answer questions. Deep learning will be replaced with LLMs. I actually think LLMs are here to stay. I don't think they'll go away.

But what I'm on is, but that's not the end technology in AI, which is the next one, and which is the next one will be an add to. There's machine learning, I think that's robust. I think there's LLMs, I think that's robust, but it's statistical in nature. And so where's the deterministic piece? Where's the knowledge piece? And is there something beyond LLMs? Look, this stuff is eight years old at this point. That's the first paper I think was 2017 when intention and this approach came to be.

Intention and transformers, the two together. Is there another one? Don't know. I suspect there is, but but we, but we don't know. The same way as in 2016, you couldn't predict, um, uh, the current LLM technology.

a comparison I would make is there's now a core technology that everyone feels very invested in. You know, when I go to San, I live in New York, and when I go to San Francisco, I joke that it's it's just a different planet. Everyone is much more optimistic about AI than I am, maybe, uh, and I look at the companies that are springing up, right? The the people that have left OpenAI to start superintelligence companies or, uh, AGI labs, they are all still betting on LLMs, right?

The core of their work is still LLMs. The idea that you can feel the AGI is a lot of people using cloud to write code and saying they can feel the AGI. Do you? Are you worried that there's not enough investment in the stuff around the edges that might supplant or augment LLMs?

No, because I think that that is when it is so so unknown. It should not be really companies that chase it. I think that academia should chase it. And I do think that there is enough AI researchers in academia who are going to be working around these topics. But when you don't make enough progress, there isn't going to be any media coverage or any other coverage. But from me talking to my friends, whether at MIT or at Illinois or at Chicago, there is.

There is work going on. I mean, like, it's just not occupying attention because the complete airwaves is LLMs only.

that's why I'm I'm asking, like, are are. Do you think that there's enough work happening? It sounds like you do, even in an environment where, you know, in America, in 2025, the pressure on universities to not bring in, you know, foreign graduates and have other kinds of, uh, academics going on, it's it seems tenuous at best. But you think that investment is still happening there.

I'm more optimistic than pessimistic. Is there some of what you described happening? Absolutely. But when I look at the number of top faculty, at least in the top, let's call it twenty engineering schools, it's not really decreasing. Is there some funding cuts? But we're talking like under 10%. It's not like it's massive. Other areas that are not the hard sciences, by hard sciences, I mean physics, chemistry, math and engineering. Yes, there are much larger numbers in some of the others, but that's not where I spend my energy thinking.

So if I think about sort of physics and hard engineering, I'd say there is some cuts, but it's not that extreme. I also look at the national labs. No cuts. So hey, it looks pretty good.

yeah, I'm, I'm happy. The frontier is good. Let me just end by sort of talking about the the the near term, right? We, we've spent a lot of time talking about how things might go, how the core technology bets you're making might play out over time, whether or not the GPU is or like dark fiber, which is my favorite. I don't know if you could tell one of my favorite arguments to have. In the short to medium term, what we are seeing is a bunch of companies saying, OK, we have AI, we can just do it.

We're gonna make the job cuts. Accenture had a bunch of job cuts, Amazon had a bunch of job cuts, UPS had a bunch of job cuts. Just in the week that we're talking. If I was to be as harsh as possible about the work of your average big consulting firm, I would look at it and say, boy, a lot of that can go. You can just let the AI make those decks all day long because the the point of this contract is to let the CEO restructure the company, right?

We, we just need the gloss of external validation to make the changes that they were already gonna make, or make the layoffs that were already gonna happen, and that is McKinsey's function in the world. And boy, that's a lot cheaper and faster to do by just letting the AI make the deck that no one ever reads in the end. I feel like I see that playing out. How do you think people should react to that inside of the timeframes that you are talking about, where the real change comes?

Could there be up to 10% job displacement? I'll use that phrase. I actually believe that's going to be likely over the next couple of years. It's not thirty forty, but it is up to ten in the total US. employment pool. It is very concentrated in certain areas. Now, as you get more productive, companies are going to then hire more people, but in different places. So that was the point I was making. We are hiring more because people say, Oh, I don't need the entry level task and the AI agent can do it.

Looking at them really like, think strategically for a moment. Won't I rather take an entry level person and the AI makes them more like a ten year expert? Isn't that more useful for me than the other way around? Because then where is my talent who's going to come up with the next great product? Where is the person who's gonna be able to convince a client to deploy technology the way it should be deployed? So I think that's why I think maybe some are being short sighted.

But I also think that some of this is right now. Because if you look at the total employment numbers through some of what you talked about, I think people gorged on employment. I've used that phrase during the pandemic and the year after. So some of that is just a I don't need so many people because I went up thirty, forty, fifty, 100% from 2020 to 2023, so there is gonna be some natural correction.

Business is never completely optimized. I think in engineering terms, it's an underdamped system, so when there's a need, it goes above. Now it's got to correct, it probably is gonna go below what's needed, and then it'll kind of come to the correct equilibrium, depending upon both market demand and growth.

Do you feel like the broader market is stable enough or predictable enough at this moment in time for that natural sort of business cycle of correction to play out in a healthy way?

Look, I've held the view for many years. People talked about, look, with all the wars, with all of the uh, cyber attacks, with all these things, with interest rates. Oh my God, we got doom is coming. Uh, GDP is going to fall. I kind of held the view. Look, if I look at the demand, I think that global GDP growth near 3% looks likely. But that ignores inflation, so in real terms, we are like 5%.

I think that those two together are probably going to hold for quite some time.

I'm just curious, cause, you know, I hear from our readers who are, you know, consumers, they they they work at tech companies, they build the products. And the split between how they feel about AI and what AI is doing to the economy, and what is claimed that AI will do to the economy is as vast as any split, as I've ever experienced in covering technology.

I think because people got trained on a certain set of technologies and people who are experts hold their expertise. They don't acknowledge it, but deeply is their identity. Now suddenly they go into work and the person who's been coding a product for ten years finds that a kid coming in from college using some gen AI tools is three times faster than them. And they didn't know the code, but the AI knows the code, but they know how to use the AI.

Well, uh, you're the CEO of IBM. Is that your experience at IBM? Because what I hear from our readers is that would be great, but it's not true. It's not happening.

We took a CO tool we built ourselves. We didn't use one of the industry tools on code to help our people do software development. Within four months, the team who embraced it, which is 6,000 people, so not a tiny number, was 45% more productive just to compare, because we have 30,000 others who don't yet use that tool compared to them. So those are real numbers. We are going to grow those teams, we are not trying to cut anything.

Because if I can be that much more productive at software development, that means that we can build a lot more products, which means we can go get more market share. It doesn't mean that it's a fixed amount of work. I think the amount of work is infinite. So if we can be more productive, the calculus always is if it's that expensive to build it, is there enough margin in what I get built that it's a viable business?

So the answer is, it's cheaper to build it. Hey, I can sell it cheaper and still have great margin. Does that make sense? Oh, it does, yeah, So that's our lived experience inside, which is why I'm leaning into hiring more programmers and more tech people.

uh, Arvind, this is great. Tell people what's next for IBM. What's I? What should they be looking for?

I think, watch what we're going to do on quantum. I think that in about two to three years, you'll see some surprising results.

Well, we're gonna have to have you back on Decoder very soon as this market shakes out. And then when the quantum bad pay like that, that that's an exciting twenty four months that I want to make sure you're back for. Thank you so much for being on Decoder.

Pleasure being with you.

I'd like to thank Arvind for taking the time to join Decoder and thank you for listening. I hope you enjoyed it. If you'd like to let us know what you thought about this episode or really anything else, drop us a line. You can email us at decoder at the verge dot com. We really do read all the emails. You can also hit me up directly on Threads or BlueSky. And now is an especially great time to reach out, cause we're gathering your questions for our end of year special.

Send them in, we'll get them answered. We're also on YouTube now. You can watch full episodes at DecoderPod, the same handle is on TikTok and Instagram. It's a lot of fun over there. If you like Decoder, please share with your friends and subscribe wherever you get podcasts. Decoder is a production of The Verge and part of the Vox Media Podcast Network. The show is produced by Kate Cox and Nick Statt and edited by Ursa Wright. Our editorial director is Kevin McShane.

The Decoder music is by Breakmaster Cylinder. We'll see you next time. Support for the show comes from Engine. Running a small business means every dollar has to work hard, but if your team is still booking travel the old way, it's costing you more than you think. Engine is the fastest growing travel and spend platform in the country, built specifically for businesses like yours. Book a trip in as little as two and a half minutes, earn up to 10% back on hotels, and in 2025, Engine customers saved more than $300 million on travel with zero booking fees, no contracts, and no BS.

More than a thousand businesses join Engine every month. Join them and get $500 when your business signs up and starts traveling at engine dot com slash vox. As soon as autumn rolls around and you swap your iced lattes for pumpkin spice, that's your signal to start thinking about the holidays, the decor, the gifts, the shopping lists, but more importantly, the holiday treats, cheesecakes, cookies. Is there anything better? And when the recipe calls for cream cheese, there's only one option Philadelphia, the original cream cheese, made with fresh milk and real cream.

This holiday season, whether you're going all out or keeping it simple, make it really Philly good. Find your favorite holiday recipe at cream cheese dot com. You're either all in on this model, or maybe you're building with another one. There's either speed or is it security, or your AWS AI. Don't pick a side, pick em all. AWS AI is how.