On National Technology Day 2026, industry leaders across India’s technology sector ask a harder question: not what AI can do, but who it actually serves — and whether we are building it with enough care to find out.
Every May 11, India pauses to mark the moment in 1998 when three nuclear devices detonated beneath the Thar Desert at Pokhran, and a nation announced to the world that it could build what others said it couldn’t. That day, which also saw the maiden flight of the indigenously built Hansa-3 aircraft became National Technology Day not because the tests signalled something harder to measure: self-reliance, ambition, and the will to lead.
Twenty-eight years later, India’s technology story has moved from the desert to the data centre. The country runs the world’s largest digital public infrastructure. UPI processes over 460 million transactions. AI is moving from boardroom pilots into production systems that touch real lives. And yet the spirit of this year’s theme — Responsible Innovation for Inclusive Growth — suggests that the harder work is still ahead.
The question in 2026 is not whether India can build. It clearly can. The question is whether what it builds reaches everyone, and whether the foundations beneath it are strong enough to hold.
The real stakes of India’s AI bet
Industry leaders have articulated what is actually at stake more directly than anybody else, Dr. Amit Sheth, Founding Director of the Indian AI Research Organization (IAIRO) and Professor of Computer Science and Engineering at the University of South Carolina said, “India has 490 million informal workers, farmers navigating climate uncertainty with limited intelligence systems, patients operating within fragmented healthcare infrastructure, and first-generation learners skilling up in languages most AI systems still barely understand,” Sheth noted. “As per a recent NITI Aayog report, AI could become one of the most powerful tools to bridge these gaps. This is not just a statistic about technology. It is a statement about what is at stake if we get this right.”
Sheth’s framing cuts through the market-size conversation that dominates most AI discussions. India’s AI market is projected to cross US $17 billion by 2027, and that number gets cited often. But Sheth argues that market size is the wrong measure. “The more important question that looms is how useful, trustworthy, and accessible AI becomes for real-world India.”
Sheth said. “India’s AI ambition cannot be measured only by market size or valuation. It must be measured by whether we build systems that expand capability, reduce inequality, and strengthen strategic autonomy.”
The data problem nobody wants to talk about
Behind every AI system is a data question that rarely makes the headline. How is data structured? Whose experience does it capture? What is the system optimised to do?
Kumar Vikas, EVP of Data & AI at Bounteous x Accolite, has been watching enterprises get this wrong, “AI has moved from pilots to production across enterprises at a pace that has outrun the thinking behind it, and that gap is showing up in ways that are genuinely difficult to course correct,” Vikas observed. “The most consequential decisions in any AI program are made early and quietly — in how data is structured, whose experience it captures, and what the system is asked to optimise for. Organisations getting this right are asking harder questions at the start, which means they are solving far smaller problems at the end.”
This is not an abstract concern. In India’s digital public infrastructure, from Aadhaar to ONDC to ABDM the design choices made at the foundation level determine whether a system serves a farmer in Vidarbha or only a product manager in Bengaluru.
Vikas sees cause for measured optimism. Cloud platforms, data tooling, and experience-led design frameworks have matured. Enterprises that invest early in evaluation, guardrails, and clean data are shipping with more confidence than those attempting to retrofit governance later. “The enterprises that win this next phase will be the ones building something their users and their own teams can actually trust.”
When AI moves at the speed of business, security gets left at the door
One of the more uncomfortable findings to emerge this year: most Indian organisations are deploying AI faster than their security and governance frameworks can follow.
Sharda Tickoo, Country Manager for India and SAARC at TrendAI, put the number plainly: “Our research tells us that 4 in 5 Indian organisations are deploying AI under pressure, often faster than governance and security can follow.”
The consequences of that gap are not hypothetical. As India builds critical infrastructure, digitises public services, and embeds AI into financial and healthcare systems, the surface area for failure and attack grows alongside the ambition.
“Ambitions without proper guardrails is vulnerability at scale,” Tickoo said. “Technology truly transforms when it is secure enough to be trusted, simple enough to be used, and inclusive enough to benefit all. For a nation positioning itself as a global technology leader, security cannot be an afterthought to AI advancement. It must be the foundation of it.”
This view is reinforced from the cybersecurity side of the equation. Diwakar Dayal, Managing Director and Area Vice President for India and SAARC at SentinelOne, pointed out that the threat has evolved in step with the technology itself. “Attackers today are able to move faster, scale operations more efficiently, and exploit gaps across endpoints, cloud environments, and identity systems. This is why cybersecurity can no longer be treated as an afterthought; it must be embedded into digital transformation from the outset.”
The identity problem inside the machine
As AI systems move deeper into enterprise decision-making, a governance challenge is emerging that most organisations have not yet fully confronted: who — or what — is making the decisions, and who is accountable for them?
Nitin Varma, SVP and Managing Director for India and SAARC at Saviynt, framed the scale of the issue precisely. “For every one human identity today, there are already dozens of non-human identities — machines, applications, and AI agents — interacting across systems. That fundamentally changes how organisations need to think about access, control, and trust.”
The question is no longer just about protecting data from external threats. It is about understanding the internal chain of AI-driven decisions. “If AI systems are initiating decisions, enterprises must be able to clearly answer: who or what took that action, what data was used, and who is ultimately accountable. AI has not introduced entirely new risks; it has amplified existing ones around identity and data.”
Varma prescribes to build governance into the foundation, not onto the surface. “Organisations that can balance innovation with strong identity controls, data governance, and accountability will not only move faster, but do so with confidence and trust.”
The invisible infrastructure behind AI
For all the conversation about AI models, there is an infrastructure layer that rarely gets equal attention, and without it, AI cannot function in real time. Rohit Vyas, Director of Solutions Engineering at Confluent India, argues that this layer is where India’s next phase of growth will actually be won or lost.
“India’s next phase of transformation is being shaped by more than just AI. It hinges on how quickly companies can process and act on data,” Vyas said. “From quick commerce to banking, enterprises are investing in systems that enable instant decisions, whether it’s fraud detection, dynamic pricing that responds to demand surges in real time, or delivery tracking that updates customers before they refresh the app.”
The challenge Vyas identifies is one that sits beneath most AI investment discussions. “Legacy modernisation remains one of the biggest roadblocks in the runway for AI, both in India and globally. A large part of enterprise data still sits in older systems, limiting how effectively it can be used. AI, by design, needs continuous context. It does not work well on yesterday’s data.”
The real shift, as Vyas sees it, is not just moving systems to the cloud. “Organisations are now realising that modernisation is not just about moving systems, but about re-thinking how data flows across the enterprise in the moment, as events happen. That is where the real unlock lies: in building systems that can stream, connect and act on data. In real time.”
Governance and speed are not opposites
A persistent myth in enterprise AI adoption is that moving fast requires moving loose, that building guardrails slows you down. Manish Bafna, SVP of Engineering at Responsive, rejects this framing.
“The enterprises making real progress are not the ones moving the fastest. They are the ones who have figured out that governance and speed are not opposites,” Bafna said. “When every AI output is traceable, when human oversight is built into the right points of the workflow, and when the knowledge powering your AI is continuously monitored for health and accuracy — that is when you can move fast with confidence.”
Bafna’s point is important as AI moves from generating responses to autonomously orchestrating decisions. The organisations that will lead, he argues, are not those with the most AI. “They are those with the most governed, most trusted, most actionable knowledge. That is the competitive advantage worth building.”
Physical AI and the governance gap
The conversation around governance becomes more urgent as AI moves beyond software and into the physical world. Piyush Jha, Group Vice President and Head of APAC at GlobalLogic, described the shift: “As AI moves from digital interfaces into the physical world, responsible innovation is no longer optional; it becomes foundational.”
Jha pointed to the emergence of vulnerability-discovery models that expose real risks of AI operating on legacy and mission-critical infrastructure. As physical and agentic AI systems begin interacting with real-world environments industrial equipment, healthcare devices, transport networks, the stakes change. “The convergence of software, data, and machines introduces new dimensions of risk, ranging from systemic failures to amplified vulnerabilities at scale.”
The response, Jha argued, cannot be post-deployment regulation. “Governance is not layered on after deployment, but engineered into the core through secure architectures, real-time observability, and accountable AI frameworks.”
Observability: The control India cannot skip
When hundreds of millions of people depend on digital systems, for UPI payments, government services, healthcare access, downtime is not just a technical incident. It is a social one.
Nalin Agrawal, Director – Solutions Engineering at Dynatrace, made the case for what he calls a new bi-modal world of technology operations, one where both human-led and autonomous AI-first models coexist, and where observability becomes the critical bridge between them.
“Digital adoption at scale — driven by UPI-level volumes, digital public infrastructure, and cloud-first strategies — means manual approaches can no longer keep pace,” Agrawal said. “Organisations must move towards AI-driven, self-healing systems that detect, diagnose, and resolve issues in real time.”
Ganesh Narasimhadevara, Director of Solutions Consulting at New Relic India, extended this point into its social dimension. “In a country where a single digital outage can affect hundreds of millions of people using government portals, UPI payments, or healthcare platforms, reliability is a social responsibility.”
Narasimhadevara connected observability directly to the theme of inclusive growth. “Responsible innovation means deploying AI that’s monitored, measurable and correctable — so digital infrastructure performs under peak load on festival days, on exam result days, on election days. It means ensuring that the digital experience for users in Bengaluru and Bhagalpur, Kolkata and Kanpur is held to the same standard of quality.”
His conclusion carries weight: “Growth that can’t be measured can’t be sustained, and innovation that can’t be trusted can’t truly be responsible.”
People must still come first
Amid all the discussion of AI architecture, data pipelines, and security frameworks, Vara Kumar Namburu, Co-founder and Head of R&D at Whatfix, returned to the most basic question of all: does the technology actually work for the person using it?
“In today’s enterprise world, the tools designed to enhance productivity often create complexity and digital friction. This disconnect impacts not only efficiency but also employee engagement and organisational momentum, ultimately hindering the very progress technology intends to drive,” Namburu said.
His answer centres on what Whatfix calls Userization; the idea of placing the user at the heart of every digital experience. “Instead of expecting people to adjust to systems, we build systems that adjust to people. The result is technology that is intuitive, contextual, and truly empowering.”
It is a principle that scales beyond enterprise software. If India’s AI ambition is truly oriented toward inclusive growth toward the 490 million informal workers, the farmers, the first-generation internet users in Tier 2 and Tier 3 cities then every system built must answer the same question: does this work for the person at the end of it?