What Will Define AI Winners: Former Cisco CEO John Chambers

John Chambers: AI's going to make the internet look boring. If you give up, you might as well fold the company. There are gonna be train wrecks, both of large companies and smaller companies, and that's something I think both the market and leaders have to be prepared for.

Steve Odland: Welcome, John Chambers, former executive chairman and CEO of Cisco Systems. John, thanks for coming today.

John Chambers: Steve, it's gonna be a lot of fun. We haven't done one of these in a couple years, and it was fun being with you in New York City honoring Lisa Su when she was CEO of the year. A great pick. I'd been her mentor and sponsor for nine years when she took the job at AMD. I'm just very proud of how she did.

Steve Odland: And she's awesome. But John, people don't realize you almost invented the internet-- i'm being only a little bit silly-- through Cisco Systems back when you were CEO beginning in 1995.

 You've been here as part of AI from the very beginning and into today, and then you've led start-ups. I just wanna hear all of this great experience as it relates to first the internet and then now leading into AI.

John Chambers: It'd be my honor. Interrupt me at any time. I'll do a line of thought. If your listeners agree with everything you and I cover today, Steve, number one, I'll be boring, and number two I'm not gonna do that. I wanna make them uncomfortable. I really wanna challenge it.

I didn't have a front row seat in these changes. I was in the middle of the playing field. And so with the internet, when we said at Cisco, "It's gonna change the way you work, live, learn, and play," people said, "You're moving around zeros and ones. Not gonna happen." But we became the most valuable company in the world. We used the internet capability across every major function.

Our profitability was much higher than everybody else's because of how we not only used the internet technology, but how we changed the business process, how we ran our company at speeds others could not match, how we could do M&A to complement our architecture, et cetera. Then I bet on AI about 10 years ago, and very candidly, at first I thought maybe I'd missed it.

I bet on AI and cybersecurity connected to AI, and people couldn't even spell it. And I really mean that only half teasing. We forget how much it's been a four or five-year phenomena to the current date. But it was the same game except, in my opinion, that what I've used from the beginning, five times the speed and three times the outcome.

I think that five times the speed is gonna be right. Three times the outcome is too conservative. It's gonna be five times the outcome. It would disrupt every company, every industry, and we make our predictions each year on the AI segment for the next year. And about the time I think I'm being extremely aggressive and people say, "You're not even close on six or seven of the transitions helping the next year," and here we are eight or nine months into 2026, and they're almost all have already happened.

I focus on market transitions driven by changes in technology which require changes in business process. If you implement technology and don't change your business process, guess what? Not much productivity. So that'd be my summary.

Steve Odland: You were CEO of Cisco Systems back, what, 30 years ago, when it was one of the most valuable companies ever for one thing. But secondly, it was the basis, the foundation of everything that's happened since in the digital transformation and the development of AI.

John Chambers: It was in many ways. We were fortunate that our competitors stayed focused on products and technology. We focused on products working together and architectures, data, voice, video, disrupting whole industries. And we did 180 acquisitions to complement our development engine at the time most people felt acquisitions would fail. Ours were, I think, still the model. As Walmart even said a year ago, it's a model not just for the tech industry, but for the industry as a whole, and some will fail within it. And then we strategically partnered, and we moved with other industry peers or key customers in ways that others did not.

And the takeaway is 40% to 70% market share and leadership in the business transformation. We were very fortunate, but I built a great team and they executed extremely well.

Steve Odland: And you talk about having been involved with AI for over a decade but really it's been three decades, John, because everything that Cisco did enabled everything that came afterwards, right? In a way, you were there at the start.

John Chambers: Agree. And I would actually say there have been two major technology eras. We talk about mainframes, minicomputers, PC, the internet, the cloud, et cetera. On it social media and then AI, but I think the two big ones are purely the internet and AI.

They do have complementary type of concepts. I've got 24 start-ups now. Eight are cybersecurity. All 24 are AI-driven, including a company like Now, which is an AI company at the heart of the network that is completely autonomous, great security built into it, much lower total cost of ownership of the transition, and AI at the heart of everything they do on it.

So taking those concepts, bringing into the start-up world, mentality of number one or number two or don't compete, and building those into start-ups is a rush. It's just really fun.

Steve Odland: Yeah. 24 start-ups. Wow. And They are successful start-ups. You've got some unicorns and decacorns in there, John.

John Chambers: We all have been very fortunate, but I've seen every movie, Steve, and after doing 180 acquisitions, if I can't pick out good companies to invest in based upon the strategy, the vision, the CEO, what customers say about them, I would have personally been disappointed. I am a risk taker on the approach, so I expect a certain amount to fail.

Two have failed out of the 24. A couple more than that will occur. But I did just for fun on the AI models, I asked what are the mathematical odds that-- it looks like we're gonna have 12 unicorns, maybe 13 or 14 in the not too distant future, and two decacorns already in Bloom and Rubrik in terms of investments.

What are the mathematical odds of that overall? And it came out with a number that I didn't even know what it meant. It's 10 to the 15th factor. So I don't think it's that high a number, but it is unusual where only one out of 1,000 start-ups get to be unicorns, and those that become unicorns, only 10% get to be decacorns.

So you do the math. And then if you combine it with being the CEO and long-term CEO of the most valuable company in the world from $70 million in sales to $48 billion, you hit numbers that are unusual. And what it says, right spot, right time. I am lucky. Luck is partially when opportunity meets preparation.

And fortunate to live at this time in life, and I think AI's going to make the internet look boring. So excited about the next chapter.

Steve Odland: Let's talk about that because, you've seen it all. You've been there from the beginning through all of these things. You've demonstrated huge success both in the companies you've run as well as your 24 investments. There's digital evolution that's been going on for a long time.

Remember when we started, there were deck riders and it was all big iron and it went from there to, they started dropping these little beige boxes in the middle of departments, and people would tiptoe around them, saying, "What are those things, and what do you do with them?"

You've seen it all. You said this does feel different. Talk to us about how it feels different.

John Chambers: I think the most important one for your listeners to understand is the speed is going to be undeniable. I think people are beginning to get a feeling for it, but the well-run companies began to implement their vision and strategy on AI two and three years ago.

And if a company isn't well into its implementation of AI, focused on key areas and producing good results, and by the end of this year the majority of its employees using AI in some ways in their business every day, they're really behind in terms of the direction. So I think that speed is important for people to understand, including how fast AI changes.

Steve, if you were in The Conference Board and you and I were interviewing some start-ups and one of my start-ups would say, ""John, I've got a new strategy I want to use this year," I would say to them, "If you're changing your strategy every year, you don't have a strategy. You're chasing the new shiny object."

That changed five years ago. Now if the CEOs don't evolve their strategy dramatic-- both of the start-ups, and you know where I'm going with this, of the bigger companies-- every nine to 12 months, they're behind. And so that speed of change where what used to occur in five years occurs in one, and the outcomes that used to take you much faster than pre-internet-- in other words, the internet was three times the speed of traditional business, but the outcomes that even with the internet that have taken place over the last couple decades, now will be at five, five times the results as well.

And so I think that's what people are missing. I think the CEOs who get it are beginning to really make a difference. Unfortunately, that bell-shaped curve you and I learned in business school and in our practices, where if you're in the middle you'd have a good future, solid returns, maybe two to five times valuation.

The far right, maybe 15% would do great, far left 15% would fail. That's getting flattened even with the $1 trillion valuation companies. I think there are now a dozen of them. And as it flattens, many more are gonna fail but the ones who are successful are gonna be obscenely successful in terms of the financials.

There are gonna be train wrecks, both of large companies and smaller companies, and that's something I think both the market and leaders have to be prepared for, and the key is how do you avoid them.

Steve Odland: A lot of that can be applied just to the AI industry, if you think of it that way, the tech industry. But the vast majority of companies are not actually in the industry, they're users of the technology. Your point, I think, is a really important one, which is speed. Now, what do you do if you feel like you're behind? 'Cause every CEO I talk to said, ""Ah, I just, I don't know. I don't know how to benchmark it, but I feel like I'm behind."

Do they just give up or is it a matter of pedaling faster or is there something else they should be doing?

John Chambers: And a great setup question because you know it's well between those two scenarios. If you give up, you might as well fold the company. You stay stagnant in this world, you're going to be roadkill, and the changes that your peers will go through, both the small ones and the big ones, will be so fast that after a couple years, quote, "giving up," if you're a Fortune 500 company, you won't even be on that list.

And so that's not an option. The second is you need to think about it differently in terms of your organization structure. What do you do? If you think of this traditional CIO, she or he was up the vertical stack. Your chief security officer went horizontal. When you think about AI, you've got to think about the CEO being the champion.

You might put the responsibility for strategy into a chief AI czar, or you might ask the CIO, chief information officer, to help lead that from a technology perspective. Or you might have your business unit leads, functional leads lead it, like CFO and finance, engineering and engineering, head of sales and that group. Or you might put a COO over top of it.

But all these groups have to work together, and the CIO has to have a common data architecture and an architecture for how they're going to use AI. It can't be a whole bunch of separate products. Same thing is true on cybersecurity. Your members of The Conference Board are going to see next year most complex cybersecurity docs ever on it, and if they aren't well ahead of thinking how they're going to handle the cybersecurity for agentic agents, both ones that they do themselves for productivity and the numbers that come in externally, they're already in trouble.

One of my companies, PinDrop, which did the leadership in deepfakes, does it for Zoom, does it for Cisco, Microsoft, and others in terms of is that the person you're talking to the right person? Is it a machine? Is it a combination of the two, etc.? You're going to see the same thing occur with third-party agentic AI agents and you saw with Huggy Face the other day, just 1,000 agents literally brought it down.

People think that's just the exception. It was a unique technology issue. Out of PinDrop's 100 customers, they did a formal investigation with and trying to understand how many third-party agents they had seen in the last two months, out of 100 customers, they had seen 750,000 unique AI agents, third party.

And so the same scenario applies there. Most of them were justified, and most of them, the companies wanted it in their environment. But the ones that aren't who they say they are, that are in there for the wrong intent and the wrong action, you've got to block and prevent them from really accomplishing challenges in your environment, like the US FBI, the Australian government, and Hugging Face as well.

So same scenario when you think about security on that, and Pindrop has a chance to really lead this in terms of these third-party agents. It will be about identity verification. It will be about intent, what are you coming in to do, and whether or not that's justified and they wanted to do it. And third, what is the action?

Except it's got to be at AI speed and close it down or prevent it if you get outside any one of those three areas where you should be implementing. So that speed comes back again and again, Steve.

Steve Odland: Yeah. No, it does. You may disagree with me, but the other way this feels different, and you touched on it a bit ago, but the early phases of the digital evolution was more hardware based, you're trying to get the right machinery in place and that sort of thing. A CTO and CIO need to focus on that. And then you had the cloud, and nobody knew where the heck that was anyway, but that was out there.

And again, it was more the IT department. AI is all about the business, and in a multi-constituent world-- customers, employees, owners, community-- you're trying to use AI to serve your constituents. So yeah, you've got the CIO, CTOs, they're involved from an implementation of the tool, but this is different because in order to affect it, you have to touch every work process. It goes well beyond IT implementation

John Chambers: Absolutely does, and well beyond just data organization on it, and it has to grow across every function together, not in independent silos. It's got to be owned by the CEO. That makes CEOs uncomfortable. But even if you're not into, quote, "AI" as something that excites you, you've got to have working knowledge of it as a CEO, and then you've got to build an organization that can help you drive it through your company.

It can't be something buried in the bowels of an innovation officer or a direct report to the CIO or the CIO him or herself. It needs to be owned by the functional leads, as we said earlier, and the top ownership needs to be the CEO.

Steve Odland: Yeah. You've Gone from a period of time when CEOs couldn't spell AI, as you said, and that wasn't that long ago, to holy cow, we've got to invest like crazy here. We can't dump enough money into it fast enough. And then all of a sudden people are going, "Where's the ROI?" Is it too early to be asking for some ROI on this investment?

John Chambers: No. I Think it was John Doerr that said if it's important, measure it and quantify it. And I admire John Doerr of Kleiner Perkins in many ways, a great leader here in the Valley and a great friend.

But basically, you should be looking at the productivity on it, and you should also be looking at new revenue generation because of it. And at a session with CIOs the other day, of 45 CIOs at a location of one of the venture capitalists and who said what was off the record, but the total I can share out of the people in the room.

Now, these were leading-edge CIOs. They get together twice a year at the sponsorship of the VC, and the VC gets to run a couple of their start-ups by them as well. But in going around the room, the average productivity increase by the end of this year, in other words, what their run rate is, will be about 15%.

It was spread between 10 and 20%, depending on the group, and only one of the CIO's productivity increase for their company will exit the year at a run rate below 10%. Now, these are leading-edge CIOs. So key takeaway, for the leading-edge companies and those that are leading regardless of industry, you better be well along the way in productivity.

Secondly, it isn't just about how do you do more with less, it's how do you use resources and move resources to where additional revenue's going to come from, new revenue streams, market adjacencies. And that's what all my start-ups are doing. In fact, Uniphore, which is one of the hardest start-ups I have, going for an enterprise AI company that runs architecturally much like Cisco did in terms of productivity into every function their run rate's going to go from about $200 million annual recurring revenue this last year to $500 million. And it's accelerating, and that's 150% growth.

Guess what their headcount increase has done in the last two and a half years? Down 17 people. Now they use that headcount productivity, and they'll move it into market adjacencies. Raj out of FedEx Express, who's a great friend and he really gets it in terms of what Fred Smith started with a great CIO in Rob Carter, and then Raj, as the successor to Fred.

He looks at it more for FedEx Express, how they use this productivity to generate dramatically more revenue and move into adjacencies. And I think that's the other transition you're seeing occur, Steve, that in your membership as an example, probably the CEOs were first focused on, "Am I gonna be able to get productivity increases?"

And if you aren't already getting that in terms of areas like programming, customer success, et cetera, coding, you're already in trouble. And they'll begin to go across the whole gamut in terms of the opportunity. But now they're looking at new revenue generation, new markets that they enter, and the integration of that.

The race is not only on. Most companies by the end of this year will have a lot of their employees, in fact, the majority of their employees, starting to already use AI. And most people, when I said that in December of last year, they said, "John, not gonna happen." In fact, all the models I asked the question of gave it less than a 10% probability of being even close to right. That gives you an idea of how fast this market moves.

Steve Odland: So what I hear you saying is that if you're a CEO today, you can't just focus on either innovation or productivity or, in other words, revenue. It's all of the above. It's how do you use these AI tools in every aspect of your business in order to obviously get more productive, so therefore you can redeploy resources and drive higher levels of growth, higher levels of customer satisfaction.

It's all of the above. These are tools to enable all of that, right?

John Chambers: Yes. And just like with the internet at Cisco, where we used the internet to redefine supply chain using outside third-party contractors as though they were our own manufacturing site, and we invented outsourcing, unfortunately, of high-end technology there.

We could close our books in 24 hours any time we wanted. That's great for the CFO, but the reason Larry Carter and other CFOs who worked for me did that was so we could run the business better, and we could see to the day how was the business running in every function. And we could see problems develop as opposed to wait for an end of quarter and say, "Dang, we got surprised," when it's too late to move.

And technical services is an example where Joe Pinto did an amazing job of integrating all this capability and constantly growing 60%-plus growth for the decade that he ran technical services for me with almost no increase in head count because of productivity and direction.

So every function, and then the real benefit is when you change the process of the whole company. True in the internet, true with AI in my opinion.

Steve Odland: But then you've got the speed issue, and you've got how fast these things are changing. So it seems to me that you've got to have a ton-- I'm not going to put a number on it-- but a ton of experiments going on all the time. And then you've got to have somebody sorting it through because you can't just throw your whole business model up in the air forever. You do have to anchor and implement some of these tools, but you've got to stay ahead on the experimentation, don't you?

John Chambers: You do, but you've got to have a replicatable process. Steve, you and I talked about it in the last session. One of the things that I did, I think many people were surprised, I'm dyslexic. And it Has weaknesses, but it forces me to go A, B, Z. So what I learned early on in life is develop playbooks that I can modify as I go, but move with tremendous speed.

So you've got to have these experiments under a playbook in a process, common data architecture, cybersecurity to really work, that then you can run with tremendous speed. So as you begin to think about not experimentation, but as you begin to think about pilots, because we're past the experimentation stage in what AI means.

But as you begin to think about running several pilots at a time, you want to think about what are the industry's top payback areas, such as coding, customer success, etc., but realize it's in every function now. It goes deep into legal. It goes deep into government affairs. It goes deep into supply chain.

And so you better begin to think about projects that you're testing the waters on and then growing, and you better track them with metrics to see if you get the results, and hold people accountable for both either the increased revenue or the productivity, and use it to free up resources. So it's a process that is moving at speed and outcomes that are, I think, only going to accelerate.

Steve Odland: Okay, so there's this need for speed, but at the same time, you can't just do this willy-nilly because there's risks around all of this from a security, governance, trust, execution. How do you balance all of that and still get the speed?

John Chambers: I think you've got to be realistic where there are valid concerns.

And if you remember back to the internet era, the data I remember, and we tend to remember favorable data, but in the mid to late '90s-- and it was the objective of President Clinton, he got the internet, how it changed the economy of the US and drive up GDP growth and productivity, etc.

And he was absolutely right, and I was honored to be at the White House stage, scared to death, announcing that area with him, with he and Vice President Gore, who, quote, "invented the internet," both of them beside me. And it was spectacular because of his buy-in and the direction. I think you're now gonna see that productivity number that peaked about 2% in the internet era go to as high as for the nation of 3% to 5% being a given.

That has huge implications, standard of living and otherwise, and all the way down through education as well. So that's a little bit how I'm seeing it evolve. I know I'm on the bleeding edge at times, but even that projection is looking something that is more likely to be a given as opposed to a long-term projection that people are saying, no, never gonna happen.

Steve Odland: Okay, final question. Look forward a decade and what is going to differentiate the successful companies from the non-successful companies? Is it simply speed? Is it brilliance around implementation? But what's going to separate the winners from losers?

John Chambers: Okay. I can give you my cut based upon my experience, and remember part of my experience isn't just the successes which you were so kind to mention, it was the mistakes I made because you learn more from your mistakes, and I've made a lot of them.

So what's going to differentiate? The first one is the companies who are in a leadership position 10 years from now or even as close as the end of this decade will be those that are constantly driving innovation, using AI not as a tool but as a part of their transition of technology and business together, that build into that cybersecurity capabilities, that create a culture and empowerment of how do they structure the organization and their workflows to the various elements of their functions in a way that gets the productivity.

A small company, like I mentioned PinDrop earlier, has 18 different workflows across the company that they focused on AI, six of which are in engineering, and they're using their productivity to fund all their new movements into deep fakes and third party agentic AI protection. You will see a culture which Warren Buffett talked about but most people forgot about, a culture in the company that accepts and learns to be comfortable with this speed of change. And you'll see the company focus on how do they build, using my words, process and playbooks to run at a fast speed.

That's what I think the winners and losers will be like. The majority of companies will not exist in anywhere near their leadership they have today 10 years from now. There'll be new movers and shakers, but there'll be some that just really evolve into tremendous powerhouses. So this is one you better change or you get left behind, and if you're going to change, you've got to do it with a replicatable process using AI and business process change together for your goals.

Steve Odland: Okay, we're going to leave it there. John Chambers, thanks for being with us today and sharing all your wisdom. And thanks to all of you for listening to C-Suite Perspectives. I'm Steve Odland, and this series has been brought to you by The Conference Board.