India’s AI Ambition Faces Its Hardest Test in MSMEs

News

1. India’s AI Ambition Faces Its Hardest Test in MSMEs

2. Why AI Still Cannot Match the Human Brain

3. Why Sovereign AI Demands More Than Just Data Centers and Chips

4. How India Could Leapfrog to AI-Native Factories

5. AI Is More Than GPUs, Leaders Say, as India Bets on Full-Stack Compute

6. How Hospitals Are Turning to AI to Win the Race Against Time

Artificial intelligence may be advancing at speed, but its adoption across India’s small and medium enterprises (SMEs) remains uneven, industry leaders said at the India AI Impact Summit, raising questions about whether the country’s AI story can extend beyond elite sectors.

At a session titled AI at Scale: Driving Adoption, Productivity and Market Access for Indian SMEs & Startups, executives debated what it would take to expand AI adoption across small businesses while ensuring safety, accountability and alignment with the summit’s framework of “People, Planet and Progress.”

The underlying question was urgent: can India’s AI story be built not just on breakthroughs, but on broad-based adoption across its startup and MSME ecosystem, the backbone of the economy?

Adoption is Uneven

Opening the discussion, Navin Bishnoi of Marvell Technology India and India Electronics and Semiconductor Association (IESA) laid out the numbers.

“In professions like software development and engineering, AI usage is very high, nearly 70% say they use some form of AI. But in financial advisory, it’s around 20%. For accountants and auditors, around 36%. So adoption varies widely.”

The divergence is stark. In some sectors, AI is embedded in daily workflows. In others, it remains peripheral.

He added: “Nearly 50% of content generated today globally is AI-generated, including journalism and other forms of writing. So AI has entered at scale in some areas.”

But that scale is selective.

For India’s startups and MSMEs, the path is less clear. Can adoption be structured around the three pillars of people, planet and progress, or will it remain concentrated in already-digitised sectors?

The Deep Tech Dilemma

If adoption is uneven, deep tech ambition is even more challenging. 

“There is no Indian deep tech company today selling at a global scale in semiconductors or AI models,” said Anand Kamannavar of Applied Materials Inc. “It’s not a lack of talent, capital, or resources. There is enough money. The problem is mindset.”

He argued that India often applies SaaS metrics to deep tech businesses, expecting quick pivots, rapid revenue and fast IPO pathways. Deep tech does not operate on that timeline.

“It takes 9–18 months to build a chip. You can’t pivot in two weeks.”

In ecosystems such as South Korea and Taiwan, he noted, the focus is on solving high-value problems and building strong intellectual property over time rather than pursuing short-term exits.

Deep tech, particularly in semiconductors and AI hardware, demands patience and foresight. “We must identify bottlenecks five to ten years ahead and start building now,” he said. The payoff may be delayed, but transformative.

Cloud AI vs Edge AI

Jyothis Indirabhai of NetraSemi placed the discussion in a broader arc.

“To me, AI is still in the first phase of its journey. Probably the first of four or five phases.”

Today’s AI landscape, he argued, is largely cloud-centric, dominated by large language models and high-cost deployments in server farms. It is centralised, energy-intensive and dependent on transmitting vast volumes of data to the cloud.

“It’s a good initiative and beneficial to the world. But is that the future? And more importantly, is that India’s opportunity?”

India can participate in cloud AI through services and foundation model work. But through the lens of people, planet and progress, the larger opportunity may lie at the edge.

People are increasingly concerned about privacy. They want instant responses. They do not always want their data travelling across the globe.

Edge AI, built on optimised devices and domain-specific platforms, could address practical challenges from waste management and surveillance optimisation to data protection.

“These platforms don’t yet exist at scale,” he said. “That’s where startups and SMEs have huge opportunities. Even today, perhaps less than 5% of this space is captured.”

Trust Is Not Optional

If scale is the ambition, trust is the precondition. 

Preet Yadav of NXP Semiconductors brought the conversation back to accountability, especially in high-stakes sectors like automotive and healthcare.

“If you buy a toy car and it fails, you replace it. But if your car’s airbag fails, it’s life-threatening.”

Automotive AI must function reliably from day one to the end of life. The same is true in healthcare.

“If AI misses even one critical diagnosis, you can’t just say, ‘AI goofed up.’ There is a human life behind it.”

Bias, too, remains under-addressed. Biased datasets lead to biased outcomes. If gender or societal inequities are embedded in data, AI systems will amplify them.

“Efficiency matters. Speed matters. But if it’s not trustworthy, we don’t need it.”

Responsible AI, he emphasized, cannot be built in isolation. Government, industry bodies like IESA, SMEs, and individuals must collaborate to ensure fairness, reliability, and accountability.

Manufacturing Blind Spot

Sundeep Gupta of Alphawave SEMI (Qualcomm) pointed to a massive but under-discussed opportunity: manufacturing.

“When we think of AI, we are blinded by big names like ChatGPT and Claude,” he said. But AI deployment in manufacturing is projected to grow at a 40% CAGR until 2030.

The real barrier is awareness. Many family-owned businesses are unaware of how AI can improve quality control, predict machine failure, or optimize operations through IoT-enabled monitoring.

Skilling gaps persist. Infrastructure remains expensive. Even where government schemes exist, shared AI infrastructure, and GPU access, awareness is low.

The opportunity is clear. The pipeline to adoption is not.

Scaling AI, Democratising Access

As the session concluded, one theme cut across all perspectives: AI at scale is not just about building bigger models. It’s about building broader participation.

For Indian SMEs and startups, the opportunity lies in identifying domain-specific gaps, at the edge, in manufacturing, in deep tech hardware, and in trusted AI systems.

The real test for India’s AI ecosystem will not be how many models it trains, but how many businesses it transforms, the speakers said.