The shift from CPU to GPU to NPU-based computing is redefining today’s enterprise endpoints. In an exclusive conversation, Navdeep Narula from Ingram Micro discusses with Saba Aafreen from EFY about what is driving this AI PC adoption and why ecosystem readiness matters as much as hardware innovation in the next phase of change.
Q. Could you briefly walk us through your journey in the enterprise technology and endpoint ecosystem?
A. I work with Ingram Micro as Executive Director for Client and Endpoint Solutions, looking after a broad portfolio that includes PCs, laptops, mobiles, tablets, printers, peripherals, and related enterprise technologies. I have spent close to three decades in the technology industry, largely focused on enterprise devices, endpoints, and the ecosystems built around them. Over the years, the role has evolved alongside the industry, but the core focus has remained on enterprise computing and endpoint technologies.
Q. How would you define Ingram Micro’s global and India scale today in terms of business and ecosystem reach?
A. Ingram Micro is one of the leading global technology distribution companies, with operations across multiple markets worldwide. In India, we have been present for nearly three decades, having entered the market in 1996, and today it is among our largest markets globally, after the United States. Our portfolio spans endpoint devices alongside deeper IT infrastructure including servers, storage, networking, cloud, software, and related services. Over time, the business has evolved beyond distribution to a broader technology ecosystem that connects vendors, partners, and enterprise customers across the value chain.
Q. How has Ingram Micro evolved from a distributor into a platform-led technology ecosystem?
A. Over the past four to five years, Ingram Micro has been evolving from a traditional distribution business into a more platform-driven technology ecosystem. The shift has been largely driven by changing customer and partner expectations around self-service, real-time visibility, and data-driven decision-making. Today, the focus extends beyond supply chain and financial services toward providing better forecasting, business insights, and a more seamless transactional experience through platforms like Xvantage, the company’s AI-driven B2B ecosystem.
Q. Beyond logistics, what does your operational backbone look like today in terms of services and capabilities?
A. In India, Ingram Micro has built a broad operational network over the past three decades that goes well beyond warehousing and hardware movement. Alongside distribution infrastructure and regional teams, the company today offers a range of value-added services including training, marketing support, financial and leasing solutions, commissioning, installation, and project management. The ecosystem also includes demand generation, partner enablement, and IT asset disposal services, reflecting the industry’s shift toward more integrated technology and lifecycle support models.
Q. How early do you detect major technology shifts through your position in the ecosystem?
A. From an enterprise standpoint, we are quite close to the customer ecosystem, which helps us sense technology transition trends fairly early. At the same time, as a partner-focused organisation, our go-to-market model is largely driven by reseller and channel interactions rather than direct end-customer engagement. So while we do pick up signals from the ecosystem, OEMs and vendor principals often have a wider and more direct view across the value chain, including system integrators, partners, and end users, and are equally, if not more, positioned to read these transitions.
Q. Where is demand currently strongest across cloud, infrastructure, and endpoint portfolios?
A. Enterprise demand is currently strongest in infrastructure, driven by rapid growth in AI workloads, hyperscaler expansion, data centres, and GPU-led computing, while cloud continues steady growth and endpoints remain a large but more stable and predictable segment. The key structural shift is the rise of AI PCs and NPU-based computing, marking a transition from CPU to GPU to NPU architectures and a gradual move of AI workloads toward the edge, supported by ecosystem players like Microsoft, Intel, AMD, and Qualcomm, alongside Windows 11 and Copilot-driven refresh cycles. Adoption is currently led by large enterprises in the IT, ITES, and education-focused segments, while SMBs lag due to limited readiness and unclear use cases. System integrators play a key enabling role as ecosystem maturity continues to evolve.
Q. How do CPU, GPU, and NPU roles differ in modern AI workloads?
A. In modern AI workloads, CPUs handle general-purpose computing and orchestration, GPUs are used for high-performance parallel processing and training-intensive workloads, while NPUs are increasingly being optimised for low power, on-device AI inference and real-time use cases. Together, they enable a hybrid computing model in which workloads are distributed across cloud and endpoint environments based on performance, latency, and efficiency requirements.
Q. Is the industry over-focused on hardware metrics like TOPS versus real performance?
A. There are broadly three ways to look at this:
- First: Ecosystem push is very strong, with industry players collectively driving momentum around endpoint AI and AI PCs to ensure they do not miss the transition and to actively shape market demand and visibility.
- Second: Enterprise consumption remains largely hybrid, with workloads still skewed toward cloud and core infrastructure, meaning value realisation at the endpoint is still moderate.
- Third: ROI maturity is still evolving and remains relatively low for endpoint AI deployments, as scalable and standardised use cases are still in the early stages.
Q. What misconceptions exist around AI PCs and enterprise AI adoption?
A. There are no major misconceptions around AI PCs in enterprises today. Most organisations are still in a testing and evaluation phase rather than making large-scale commitments, and they are not expecting an immediate shift of workloads from cloud to endpoint or investing heavily in endpoint AI without clear use cases. The approach remains measured and grounded in real-world deployment realities. At the same time, ecosystem fragmentation across players such as Intel, AMD, Qualcomm, NVIDIA, and Apple is not new. Enterprises are already used to evaluating multiple platforms and continue to make decisions based on workload fit and tangible value rather than broader ecosystem narratives.
Q. Is hardware ahead of software in enabling NPU-based applications today?
A. The simple answer is yes, hardware is clearly ahead of software today. The hardware ecosystem has become highly organised and synchronised, with a clear focus on AI PCs and NPU-based systems as the next major growth opportunity. It is also relatively consolidated, with a limited set of global players driving this transition in a coordinated way.
In contrast, the software ecosystem remains far more fragmented. While players like Microsoft are leading from the front, there are still many ISVs and developers for whom standardisation and maturity remain uneven. As a result, hardware is currently ahead, and the full potential of endpoint AI will depend on the software ecosystem catching up with more compelling and widely adopted local AI applications.
Q. Where does AI PC currently sit within enterprise investment priorities?
A. AI PCs are still not at the top of enterprise investment priorities. While there is a growing enterprise AI software ecosystem, endpoint and edge AI remains early in its maturity, particularly in terms of standardised applications and clearly defined use cases. Even though NPU-powered PCs are now available, the main gap is still on the software and application side, with use cases evolving across areas like generative AI, agentic AI, and recognition-based workloads. At this stage, hardware readiness is strong, but the ecosystem is still catching up. Over the next few years, NPU-based AI is expected to augment cloud-based AI rather than replace it, with a gradual shift toward more hybrid workloads where endpoints take on selective, low-latency processing while cloud continues to handle heavier compute.
Q. Is enterprise AI adoption still infrastructure first, or shifting toward endpoints?
A. From an enterprise standpoint, there has been no fundamental shift yet, with organisations continuing to remain infrastructure first rather than endpoint first, and this is unlikely to change in the near term. Most enterprises are still operating in a hybrid model where cloud handles large-scale compute-intensive workloads and on-premises infrastructure supports sensitive or controlled environments, while endpoints remain an emerging layer in the AI stack. Although hardware readiness for AI PCs is improving, a true shift toward endpoint-driven models will depend on the maturity of localised applications and use cases, which are still evolving.
Q. Is AI PC momentum driven more by enterprise demand or ecosystem-led refresh cycles?
A. It is both genuine enterprise interest and ecosystem-driven push, but the stronger force today is industry-led acceleration, driven by FOMO around not missing the endpoint AI shift. While enterprises are beginning to evaluate AI PCs, much of the momentum is being shaped by vendors actively driving use cases to keep the endpoint central in the AI stack rather than it being absorbed into a cloud-only model. At the same time, cloud alone is insufficient, since real workloads still require local processing and applications, making edge capabilities essential. This is why players like Microsoft, Intel, AMD, Qualcomm, HP, Dell, and Apple are collectively driving the transition, as endpoint AI is being actively accelerated in the next growth cycle.
Q. Are there differences in AI adoption between Indian and global enterprises?
A. I do not think that is the case. The readiness of Indian enterprises is as high as their global counterparts. I think it is the SMB and small- and medium-business segment that is possibly not adopting or embracing AI as much. But as far as enterprises are concerned, the acceptance is absolutely world-class.
Q. What bottlenecks exist in scaling AI infrastructure through partners?
A. The main bottleneck is conviction, as partners and customers are still not fully convinced due to unclear use cases and inconsistent ROI, which is keeping adoption cautious. There is also a clear readiness and capability gap across the channel, with maturity varying significantly from global system integrators to smaller resellers. To address this, we are working closely with vendors like Microsoft, Intel, AMD, and Qualcomm on training and enablement, while also supporting execution through financial structuring, solution architecture, deployment support, and lifecycle services. The aim is to simplify adoption and make AI deployment more practical and frictionless as the ecosystem matures.
Q. Is Ingram Micro evolving into a consulting-led ecosystem orchestrator?
A. Distributors like Ingram Micro now have no choice but to evolve into consultants as well. One size does not fit all. Some partners are fully capable independently, but many rely on us for presales consulting, solution architecture, system design, deployment, managed services, commissioning, installation, and ongoing management. So we have built a modular, drop-down style support model where partners can choose the specific capabilities they require. Alongside this, we provide financial leasing, CapEx-to-OpEx transition support, and IT asset disposal services, creating a complete enablement portfolio to support enterprise AI adoption and endpoint refresh journeys.
Q. Are you expanding your partner ecosystem for the AI era?
A. Partners remain our primary customers, so ecosystem expansion is a continuous priority for Ingram Micro. We actively identify and support emerging partners with strong potential but limited capital, and in some cases provide both technical and financial backing to help them scale over time. We have also expanded the ecosystem through initiatives like our cloud marketplace, which has brought in a new class of service and subscription-led partners who have gradually expanded into broader infrastructure, hardware, and endpoint solutions. Overall, partner acquisition, enablement, and ecosystem development remain central to our long-term growth strategy.
Q. How will enterprise computing evolve over the next 3 to 5 years?
A. Enterprise computing is moving through a clear architectural shift from CPU to GPU and now toward NPU-based systems, which is reshaping how computing is built and consumed across environments. Over the next three to five years, the model will remain hybrid, with endpoint NPUs complementing cloud and enterprise infrastructure rather than replacing them, and workloads distributed based on latency, data sensitivity, and compute intensity. As device capabilities improve and more intelligence moves to the edge, a greater share of AI workloads will gradually shift closer to endpoints, driven by real-time decision-making, lower latency, and secure local processing.
Q. What role will Ingram Micro play in the future AI ecosystem?
A. Our AI focus extends beyond endpoints into infrastructure, cloud, and the broader AI computing ecosystem. Globally, we have dedicated AI teams and a consulting practice that helps partners and customers design and deploy AI solutions across different environments and use cases. In India as well, we are building focused AI capabilities to enable channel partners and support enterprise adoption at scale.
At the same time, we are driving AI internally through our Xvantage platform, which uses partner behaviour, transactions, and usage data to improve recommendations and insights. Overall, AI is a key strategic pillar, and we see the shift to AI PCs creating broader opportunities across the endpoint and computing value chain.
