Ashwin Venkatesan — Executive Research Leader, HFS Research [00:22]
Hello everyone, welcome to another episode of HFS Unfiltered. I’m Ashwin. I lead the Technology Services Analyst and Advisory practice here. Recently, we launched a very interesting Horizon Report on the next-generation infrastructure services theme, which is a pretty hot topic now with everything that’s happening with AI. In that analysis, LTM has been positioned as a Horizon 3 provider. To talk about it a bit more, I have with me my good friend, Pandi, who is the EVP for Cloud, AI Platforms, and Infrastructure Services at LTM. Pandi, congratulations, and I’m very happy to have you here.
Pandiya Kumar Rajamony — Executive Vice President, LTM [00:58]
Thank you. Thank you, Ashwin and HFS, for hosting me today.
Ashwin Venkatesan — Executive Research Leader, HFS Research [01:03]
Yeah. So, Pandi, excited to have you here. I just wanted to talk about a few themes which we also picked up on in this report. Firstly, when we think of infrastructure services, it’s becoming the control plane for enterprise AI, right? There’s a whole lot happening in terms of cloud and core and edge. But really, when we speak to enterprises, it’s coming down to how everyone can drive confidence in workloads from an AI perspective, be it cost, trust, or performance. So, what are you picking up in your conversations, and how are you seeing enterprise expectations evolving from an infrastructure standpoint?
Pandiya Kumar Rajamony — Executive Vice President, LTM [01:40]
Yes, we are starting with a great question. If you look over the last decade, infrastructure has evolved from just hosting workloads to becoming a control plane for enterprises in the digital transformation world itself. This shift has happened to the level of infrastructure becoming the control plane. It started from managing workloads to orchestrating workloads between hybrid and multi-cloud environments, right? We have seen that most of the workloads were transactional until recent times, but with AI, the workload scale and complexity have changed dramatically. Today, it’s not just about CPUs. Enterprises are managing cloud-based AI services, AI accelerators, and edge systems that are adding more complexity. At the same time, all of this has to work in a coherent way. That is causing more challenges for enterprises, and that makes infrastructure, as a control plane, work in a very different way. Then you have another complexity adding into this, which is sovereignty. Today, most countries are looking for data sovereignty, and that leads to infrastructure sovereignty as well. In the past, infrastructure mainly decided where workloads should run, but now, with AI, it must also decide what kind of compute is required, how to place it, and how to orchestrate it. That’s a fundamental shift. Because of this, enterprises are expecting infrastructure to evolve in four key ways. One is that infrastructure must become AI-aware, as training and inference behave very differently. It must be more dynamic, placing workloads across cloud, data centers, or the edge. Security and governance become even more critical, especially with AI agents. Finally, cost management is very important. Tokenization and the cost of GPU infrastructure are adding more complexity when AI is deployed at scale. These four elements are becoming increasingly important for all enterprises. Now, the real question is no longer, “Can my infrastructure run AI?” The real question is, “Can my infrastructure help me scale AI safely, efficiently, and intelligently?” At LTM, we see this as a move to AI-optimized infrastructure built for scale, speed, and trust. It is not just about adding GPUs. It’s about building a smarter, more intelligent foundation that keeps enterprises in control as AI scales. That’s the real shift. In the AI era, infrastructure is no longer just an enabler. It becomes an intelligent control plane that will define who leads and who falls behind.
Ashwin Venkatesan — Executive Research Leader, HFS Research [04:42]
Well put, Pandi. In fact, as part of our analysis, we saw the new positioning that LTM has taken around this portfolio, which is Cognitive Infrastructure Services. We understand that the BlueVerse platform is right at the center of all of this. Can you highlight one or two key ways in which LTM is evolving its capabilities and platforms to address these enterprise expectations?
Pandiya Kumar Rajamony — Executive Vice President, LTM [05:09]
Sure. See, in the AI world, the winners will not just build better models. They will build smarter infrastructure to run those models. At LTM, our shift to Cognitive Infrastructure Services is built on this idea, moving from traditional services to a platform-agent operating model. So, it’s not just a name change. Fundamentally, you change the way you look at it. You change your operating model itself. We are strengthening our control plane capabilities across the lifecycle, from design to run, including infrastructure right-sizing, right placement of workloads through orchestration, AIOps, and security. At the same time, we also look at the sovereignty aspect when placing the right workloads in the right environment. We also help clients modernize their transactional workloads because it’s going to be a heterogeneous environment. I would not use the word “hybrid.” It will be heterogeneous workloads between CPUs and GPUs sitting in the same infrastructure landscape. So, current workloads have to be modernized so they can run efficiently in this heterogeneous world. A lot of focus goes into looking at current transactional workloads and modernizing them as well. That’s the reason we thought we have to make both existing legacy CPU workloads, transactional workloads, and GPU workloads managed very effectively. The cognitive function has to come into the data center. When it comes to operations, we have been talking about automation so far. In the new world, AI-native operations are not achieved by simply adding automation or layering AI on existing processes. If you do that, it will only create pockets of productivity, not holistic productivity, and it doesn’t truly transform operations. The real shift is to rethink the entire operating model: people, processes, platforms, knowledge, context, and governance, with AI at the core. This is where our BlueVerse and IRUN come into the picture. IRUN, the concept and solution that we have created, unifies infrastructure, platforms, and application operations into a single AI-native framework, enabling a move from reactive to predictive and increasingly autonomous operations. This is powered by our BlueVerse platform. The entire automation and AI infusion come from our BlueVerse platform, which I’ll explain as we go along. What we do is build enterprise context through knowledge graphs that connect system dependencies and business priorities. The knowledge fabric plays a very crucial role, and that comes from our BlueVerse platform. We are also reimagining workflows, making processes reusable, measurable, and continuously improving, creating full-stack observability. That’s very crucial in the AI world. You need complete observability across your infrastructure, applications, cloud, and business services. This full-stack observability is not built in the traditional way. It is built using AI, and that comes from our BlueVerse platform. Finally, we are evolving the role of engineers. We have traditionally developed skills in silos or through limited cross-skilling. But when you move to AI-based operations powered by full-stack observability, engineers have to think differently about how they handle issues. Their role shifts from executing tasks to supervising and governing AI agents. So, we are embarking on a huge talent transformation as well. We are also building domain-specific small language models grounded in enterprise context. Behind all of this, BlueVerse, our human-plus-AI collaborative platform, orchestrates agents, workflows, governance, and human oversight. For us, Cognitive Infrastructure Services is not just the next version of infrastructure. It is a fundamental shift to an AI-native operating model.
Ashwin Venkatesan — Executive Research Leader, HFS Research [09:54]
Yep. Yeah, absolutely. That’s exactly the operative word here, Pandi, the post-AI implementation operating model. You touched on various themes. For our audience and everyone listening, can you bring this to life by sharing one recent client example where you’ve taken them through this journey, and what value they were able to observe and realize?
Pandiya Kumar Rajamony — Executive Vice President, LTM [10:18]
Sure. First, it’s important to recognize that our clients are all at very different stages of the AI adoption lifecycle, ranging from early experimentation to scaling and increasingly moving toward AI-native operations. There are various stages, and clients are in different phases of adoption. We work with our clients across all of these stages. I’ll share two live examples where we helped clients create value. I’m sure you’ve recently heard about our landmark engagement that we announced in the market for the Central Board of Direct Taxes in India, to modernize India’s tax analytics platform under Insight 2.0. This complete solution is powered by our platform for infrastructure. What we are doing is building a secure, intelligent AI backbone that enables smarter citizen services, automated operations, and AI-driven decision-making. What is important is how we did it. Instead of overbuilding expensive GPU infrastructure, we applied our Metal-to-Intelligent framework to dynamically optimize workloads, delivering the same performance with a 37% lower GPU footprint. If I look at another example, on reimagining operations for one of our long-standing travel clients, we introduced our IRUN agentic AI model by bringing all the metadata together, putting it into our knowledge graph and knowledge fabric, and then creating a super agent that helps run operations in a very efficient way. That IRUN super agent started yielding results by bringing down operational costs by 35%. What it also helped us do was expand our engagement with the client, doubling our revenue. While we improved cost efficiency, it also helped us grow the relationship and double our revenue. That’s what we’re seeing in the market today. So, IRUN optimization is not just about optimization. It also gives us many opportunities to improve our revenue landscape with customers across these transformation journeys. If you look closely, a few themes remain very consistent: platform AI with BlueVerse to industrialize deployment, operations through IRUN, and our BlueVerse knowledge fabric that connects data, telemetry, and workflows. This is what enables enterprises to move from reactive operations to intelligent, real-time decision-making and truly create value at scale.
Ashwin Venkatesan — Executive Research Leader, HFS Research [13:15]
Yep, this is very exciting. Pandi, thank you for sharing this. We’ve been impressed by the LTM story in this space, and we’ll continue to track how it evolves. Thank you so much for taking the time to share these insights, and we wish you all the very best.
Pandiya Kumar Rajamony — Executive Vice President, LTM [13:29]
Thanks. Thank you, Ashwin. It’s a great opportunity for us to explain what we are doing with BlueVerse and IRUN. Thanks for hosting us. Thank you.
