In an interview with TimesTech, Abhishek Agarwal, President – Judge India & Global Delivery at The Judge Group, spoke about why Edge AI and semiconductor innovation are critical to India’s long-term AI ambitions. He highlighted the importance of AI systems built for Indian realities, semiconductor self-reliance, and scalable edge infrastructure that can support manufacturing, healthcare, telecom, agriculture, and smart city transformation beyond urban centres.
Read the full interview here:
TimesTech: Much of the AI conversation in India currently revolves around large language models and cloud-based AI systems. In your view, why is Edge AI becoming increasingly critical for India’s long-term AI growth story?
Abhishek: The cloud-first AI conversation is important, but it tells only part of the story. India’s real AI opportunity lies at the edge, where decisions need to happen in real time, in places where cloud connectivity is unreliable, expensive, or simply unavailable. Consider a quality control system on a factory floor in Pune, a diagnostic tool in a rural health centre in Jharkhand, or a traffic management system in a tier-two city. None of these can afford the latency or the data costs that cloud dependency introduces. Edge AI processes data where it is generated, delivers faster responses, reduces bandwidth consumption, and operates independently of network conditions. As India’s AI ambitions move from urban pilots to genuine national-scale deployment, the infrastructure that makes that scale possible is edge infrastructure. Ignoring it means building an AI ecosystem that works beautifully in Mumbai and breaks down forty kilometres outside it. That is not a growth story. That is a gap.
TimesTech: India faces unique challenges such as inconsistent connectivity, infrastructure gaps, and cost sensitivity. How can the integration of Edge AI with semiconductor innovation help address these real-world constraints more effectively than traditional cloud-first AI models?
Abhishek: The combination of Edge AI and semiconductor innovation directly addresses the three constraints you have named, and it does so in ways that cloud-first models structurally cannot. Connectivity inconsistency is neutralised when AI inference happens locally on the device rather than depending on a round trip to a data centre. Infrastructure gaps become less limiting when the hardware running AI is compact, low-power, and deployable without significant civil or electrical infrastructure. Cost sensitivity is addressed through purpose-built AI chips designed for specific workloads rather than general-purpose compute, which dramatically reduces both unit cost and energy consumption. India’s semiconductor push, through initiatives like the India Semiconductor Mission, creates the possibility of designing chips optimised for Indian deployment conditions: heat, power variability, specific language and dialect processing, and the kinds of sensor data that Indian industrial and agricultural environments generate. This is not about copying existing semiconductor roadmaps. It is about building hardware that fits the actual problem rather than retrofitting the problem to fit imported hardware assumptions.
TimesTech: With initiatives like the India Semiconductor Mission 2.0 gaining momentum, how important is semiconductor self-reliance in enabling India to build scalable and sovereign AI infrastructure?
Abhishek: Semiconductor self-reliance is not a nice-to-have in India’s AI infrastructure story. It is foundational. Every AI system, whether at the edge or in the cloud, runs on chips. As long as those chips are entirely imported, India’s AI sovereignty is conditional. Supply chain disruptions, export controls, geopolitical tensions, and vendor pricing decisions made in other countries all become variables that India cannot control. The semiconductor shortages of 2021 and 2022 demonstrated to every government and every industry globally what chip dependency actually costs when the supply chain breaks. India Semiconductor Mission 2.0 is the right directional response, but the depth of the commitment matters as much as the intent. Building genuine design and fabrication capability takes years and requires sustained policy support, talent development, and patient capital. The payoff is an AI infrastructure layer that India owns, that can be optimised for Indian requirements, and that does not carry the geopolitical vulnerability of complete import dependence. For a country with India’s scale and ambitions, that is not optional.
TimesTech: Across sectors such as manufacturing, healthcare, smart cities, telecom, and public services, where do you see the most immediate opportunities for Edge AI adoption in India, and what kind of impact can it create on the ground?
Abhishek: Manufacturing is the most immediate opportunity, and the impact case is already being proven. Predictive maintenance, visual quality inspection, and real-time process optimisation on factory floors are delivering measurable cost reductions and yield improvements at facilities that have implemented Edge AI properly. Healthcare is close behind. In a country where specialist medical expertise is concentrated in urban centres, edge-based diagnostic tools that can analyse imaging, flag anomalies, or support clinical decision-making at the point of care have the potential to change outcomes in districts where that expertise has never been physically present. Smart city applications in traffic management, public safety monitoring, and utility optimisation are progressing in cities like Surat and Pune with tangible results. Telecom networks are using edge computing to manage traffic intelligently and reduce latency in ways that improve service quality without proportional infrastructure investment. Public services, particularly in agriculture, where soil sensors and weather data processed at the edge can drive precision-farming decisions, represent a transformative opportunity that remains significantly underdeveloped relative to its potential impact on rural livelihoods.
TimesTech: From your experience in global technology consulting and enterprise transformation, what are the biggest challenges organisations face today in operationalising AI systems beyond pilot projects and proofs-of-concept?
Abhishek: The pilot-to-production gap is one of the most consistent and expensive problems in enterprise AI today, and it is rarely a technology problem at its core. Organisations run successful pilots in controlled conditions with dedicated teams, clean data, and executive attention. Then they try to scale into real operational environments where data is messy, processes are inconsistent, legacy systems were not designed for AI integration, and the teams responsible for day-to-day operations were not part of the pilot at all. The transition breaks. Three specific challenges come up repeatedly in our work. First, the data infrastructure that was adequate for a pilot becomes a serious bottleneck at scale, because the volume, variety, and governance requirements change significantly. Second, change management is consistently underinvested. AI systems change how people work, and organisations that treat adoption as a technical implementation problem rather than a human behaviour problem are consistently disappointed. Third, the internal capability to maintain, retrain, and improve AI systems after deployment is often absent, leaving organisations dependent on vendors for what should be a core operational capability. Solving the pilot-to-production problem requires treating it as an organisational transformation, not a technology deployment.
TimesTech: As India positions itself as a global digital and AI powerhouse, what strategic steps should enterprises, policymakers, and technology leaders take to ensure the country builds AI solutions that are practical, scalable, and designed for Indian realities rather than imported use cases?
Abhishek: The most important strategic shift is moving from AI adoption to AI origination. Adopting solutions designed for Western or East Asian markets and adapting them to Indian conditions is a valid short-term approach, but it does not build the capability or the competitive position that India’s ambitions require. Three specific steps matter enormously. First, invest in Indian-language AI development at the infrastructure level. India’s linguistic diversity is not a deployment challenge to be managed. It is a design requirement that shapes everything from data collection to model training to the user interface. Building AI that genuinely works across India’s language landscape is a competitive advantage in global markets as well. Second, create regulatory clarity that encourages responsible deployment without creating compliance frameworks so complex that only large incumbents can navigate them. India’s AI regulation conversation is still early enough to get this balance right. Third, build sector-specific AI talent pipelines that combine domain expertise with technical capability. A healthcare AI system built by people who understand Indian clinical realities will outperform a technically superior system built without that context. India has the domain expertise and the technical talent. The strategic task is connecting them systematically through curriculum design, industry-academia partnerships, and deployment environments that treat Indian problems as the primary design brief rather than an afterthought.
