In The Arena by TechArena - Carrier’s Arun Nandi on Early AI Adoption and the Infra Edge

Episode Date: October 28, 2025

Recorded at AI Infra Summit 2025 in Santa Clara: Carrier Chief Data & AI Officer Arun Nandi on infra as AI’s backbone, how early adopters win on ROI and speed, and what changed in the last 12–24 months.

Welcome to Tech Arena, featuring authentic discussions between tech's leading innovators and our host, Alison Klein. Now, let's step into the arena. Welcome in the arena. My name is Allison Klein. We're coming to you from the AI Infra Conference in Santa Clara, California. And I am so excited because we're with Arunani. He is the chief data and AI officer. a carrier. A Ruin, welcome to the program. Thank you, Alison. Thanks for having me. So you've been on Tech Arena before, but this is the first time in new role. Very exciting role.

Why don't you tell me about Carrier and what you're up to? Yeah, thank you. So very excited to be leading the charter for enterprise-wide data and AI. Carrier is a Fortune 200 company, very large enterprise, has a presence across many different countries and business units, and really excited to be a part of share. shaping this next chapter of that journey. And of course, no stranger to all things on your podcast as well as everything within AI. So it's great to be here. Now, Carrier, obviously, a deep manufacturing company, tons of heritage in manufacturing. When you unpack the AI

opportunity from that context, how do you see that evolving within a large manufacturer like Carrier? Yeah. So I really see the distinct set of opportunities as one, within the order to cash or SG&A ecosystem, which is typically the IT domain. You have the operational domain, which is typically all of the manufacturing spaces. You have the different pillars within manufacturing and planning and what you do there. And you have a lot you can do within AI driving digital twin robotics and the like there. And third, you have the product ecosystem, which is essentially harnessing all of the data that we get from our assets, our physical assets that are out there in people's homes,

in data centers, in commercial buildings as well, and what you do out of that and how you make those assets intelligent. So there's a lot that manufacturing companies can and are doing with AI. We've seen a couple of keynotes today from some luminaries within the space and what they're doing to further this whole journey of robotics and operational AI. You know, what's interesting, I talked to you about a year ago, And we were just talking about just gathering data and figuring out the opportunity statement of AI. And the thing that I like about AI impras, it brings together folks from the enterprise, people who are running businesses outside of the realm of Silicon Valley, outside of the hypers, and really putting it into practice into how to develop opportunity for business.

Now, obviously a lot has advanced in the last year. So where do you see enterprise adoption of AI today? Where is it happening, especially within the generative and agendas space? And how does your cross-company experience really put you in a great position to deliver discrete value for carrier? Yeah. So great event, really excited and happy to be here. Third year in a row for me to be doing a keynote here. So very happy to come back here.

I've seen this event grow and really shape into what it's been. come now. So it's been exciting to see that journey and see all of the different practitioners, the different people that are furthering this whole cycle that we're in as well participate in this event. So it's always great to come back here. I think from an enterprise adoption perspective, there's obviously a lot that we are doing. I would quote Jensen from Nvidia saying that AI is the great equalizer, right? We've seen AI companies, SaaS company which are leveraging AI with no domain experience at all, come in and deliver some compelling products and solutions

that are really revolutionizing how we address certain use cases. So I really see that as an opportunity, huge opportunity for us to leapfrog some of the barriers that existed within domains, verticals as well. And I think AI has an opportunity to transcend beyond some of those borders. When I talk to others in the industry, one of the things that we talk about is a Gentic AI. and the opportunity for agents to be working side by side with employees.

When you look at opportunity and risk with the Gentic AI, how do you think IT departments and AI officers like yourself are looking at that opportunity and thinking about integration, given all of the best practices around data security? Yeah. I think the whole concept of a digital worker has really evolved, right? So the concept of having a string of agents that perform a knowledge worker's tasks or a set of activities is something that we've, as an industry, we've evolved to appreciate now. And that has been a great development because you've come into understanding that there are now sets of tasks and processes which are completely automated through not just a series of operational tasks, but also reasoning tasks.

And this really is the art of the possible with what we've got going with Agenic. And I think Agenic AI, of course, we've really evolved the spectrum of AI from more traditional AI with machine learning to generative AI, which was all the talk a couple of years ago to now Agenic AI. And I think the ability for us to play the entire spectrum for each use case and each domain and address that through a combination of them really unlocks a lot of that value. Now, Arun, you know, one of the things that I talk to you, off-camera about, I thought was fascinating, is really how you guide your team from concepts

to pilot to actual deployment of a particular solution. And you said something interesting in that conversation of like, there are endless pilots because you're looking for value. Tell me about how you tackle that as a leader, driving innovation, but also keeping an eye on delivering an R-O-Y for the company. Funny, I heard somebody refer to it as Death by a Thousand POCs. And obviously we were a couple of weeks out from, let's say, InfoS, MIT article that came out

that talks about a lot of your pilots and POCs failing and 96% of them failing. And I think it's just the nature of the industry that we're in, that we're experimenting. And experiments should have a start and an end date. We need to really adopt this hypothesis of fail fast and scale fast. And when we have an opportunity

which results in some untapped potential that we want to scale, we have equally that same opportunity to take that as it is on us to stop a project which is not working. So I think there's a lot within that that we need to do as an industry, but I'm a big proponent of rapid prototyping

because without experimenting, you don't yet know the art of the possible and it's really hard to get to understand that if you're not trying and innovating on new concepts. Many initiatives struggle to deliver measurable business value. You just talked about that in terms of the recent report. And I don't think I've stopped talking about that report since it's been published. It comes up in every conversation.

How do you approach measuring ROI at scale and getting some other business leaders to buy into something that they may have some fear around? Yeah. So I go back to Robert Solow, the Nobel laureate. who spoke about the computer age about, let's say, 25, 30 years ago. And he commented that I hear about all of the development of this computer age except in the statistics, in neighbor data, right? And he was commenting about this maybe seven to 10 years before we started seeing

the realization of this ROI. And I think we're in that phase right now with the AI cycle as well, where we have seen the art of the possible. seen the opportunities that exist and what AI can do from a productivity lens, but with all of the experimentation, with all of the innovation and all the energy and investment that's going behind it, we're also seeing that that hasn't yet fully translated into productivity, depending on how you define productivity as a output divide by a unit of input. And today we still refer to a unit of input as a human being and the tasks that they do. But I think in future,

we're probably going to refer to it as a unit of compute potentially, which is, you know, going to take over a lot of those tasks as well. So yes, we are in the beginning of that cycle, whether this is year three of a 10-year cycle, a 15-year cycle or a 7-year cycle, I think, is anybody's guess. But we absolutely need to continue to progress from this early stage to a more mature stage, which is when we'll start to get to realize a lot of those benefits

that we're talking about. And I think in pockets, we still do realize them, is just yet to manifest itself in large statistics that we see across the board. Now, one thing that I think everybody is thinking about is AI is a machine and a data problem, but it's also a human problem to solve, which is how do we get workers to fully integrate into using all of the AI tools available

to drive more productivity and efficiency into their work? And how do you talk to other business leaders inside your company to make that adoption successful. Yeah. So I like what Satya Nadella said about this. He spoke in a small gathering last week, and I had an opportunity to be a part of that. And he said,

personal productivity is every single individual's responsibility. And I love that comment because it goes back to all of the different co-pilots that are out there and everything that we're doing. I see co-pilots as a gateway drug for us to essentially expose AI to the masses. Because what we're going to ask of, these individuals, of all of the knowledge workers, the frontline workers, etc., is to do much more sophisticated things with AI. And for them to get comfortable with a co-pilot or an assistant

or an agent that they can start playing around with on a day-to-day basis is incredibly important for us to drive that people change and that process change as we go forward into this cycle. So I really think it's on us all to look at our workflows, how do we optimize it, how do we drive productivity within it. And for the practitioners out there, continue to evangelize and advocate for the adoption of these smaller assistants, which are going to be the stepping stone to much larger things that we need to do. That's a great example of Sacha. And I think he's done such a good job at Microsoft of really walking the talk in the way that he's setting up expectations of his leaders around AI adoption. And I think that's been written about quite a bit.

The next question that I have for you is one about organizational culture and maybe Microsoft is an example of this of where do you see organizational cultures succeeding and what are the traits of an organizational culture that are going to be perfect for adoption of AI? Yeah. Having an ability to have a foresight

is anybody's dream, right? An investor's dream. If you could have the foresight into what is the art of the possible, but I think companies that succeed are the ones that are early adopters, early adopters to technology before it becomes stable stakes. And if you look at the way that we progressed across different technology cycles, if you think of the cloud, if you think of previous cycles,

big data as a previous cycle before that as well, what typically tends to happen is that the early adopters tend to get a leg up both in their ability to accelerate at speed once they made that initial investment, as well as from a cost perspective, to be able to hit the ground running when the cost haven't yet spiraled out of control. The later adopters come to the table when it is stable stakes, when it's absolutely imperative for you to invest in that technology. And that's when mature stages, you are spending a lot more on that.

The ROI curves become a lot lower as well. So I think companies that do succeed have this culture of experimenting, of having the right ideas on what to invest behind, but also being early adopters do a lot of different things. Some may fail, but that's the nature of how we will progress as a technology ecosystem, which is that the one or two that succeed out of the 10 are probably going to be able to recoup all of the investments that you've made across the board. Now, I'm going to do a right-hand turn on you because I have been talking to you about

organizational culture and the process for implementing AI, where we're going with it. But you keep coming back to AI Impra Summit, which tells me that as a data guy, you want to know about the infrastructure. Why is the underlying infrastructure development so important in this space? And why do you prioritize it as a focus? It's the backbone of everything that we do. You and I were talking about it off camera as well, which is how this summit has evolved.

There was a lot of talk about the sustainable infrastructure. And now it's all about the talk of accelerating infrastructure. The world has really moved quite a bit in these past 12 months and in the 24 months since I first attended this summit as well. Well, infrastructure is and continues to be the most important enabler for how we drive AI. Many will argue that foundation models have probably brought this AI cycle to start with ChatGPT launching to the consumers in late 2022. I would argue that the precursor to that was the revolution that we had in the infrastructure space

with really having the accelerated ability of these chip manufacturers, the GPUs, and all the progress that they've made. And I like that the AI InfraSummit brings together the intelligence that you get from the Silicon and the intelligence that you get from the data and bringing those together to drive the maximum output offer. I love that. Arun, every time I talk to you, I learned something more.

Today was no exception. I know that our listeners are going to want to talk to you. So where can they go to connect with you and continue the dialogue? Thank you for having me. And it's always a pleasure to be here. LinkedIn is the best way to get in touch with me and happy to continue to engage with folks out there. Thanks so much for being on the program.

Thank you. Thanks for joining Tech Arena. Subscribe and engage at our website, Techorina. All content is copyright by TechRena. Thank you.