Hyperscale alone won't work for India: HP's Ipsita Dasgupta backs LLM–SLM hybrid strategy 

Ipsita Dasgupta, Senior Vice President and Managing Director of HP India, Bangladesh, and Sri Lanka, speaking at the AI Impact Summit 2026, said India must adopt a balanced approach to artificial intelligence that combines hyperscale capabilities with smaller, distributed models.

I'm actually going to create a visualization of some of these concepts and why this hybrid of LLM and SLM is going to be really important and how we can't just depend on the hyperscale model,” she said.

LLMs, or large language models, are powerful AI systems that require massive data centres and computing capacity, while SLMs, or small language models, are lighter, more efficient versions that can run on local devices or edge infrastructure.

She noted that while large-scale AI infrastructure is important, the investment required is enormous.

“One gigawatt of AI data center capacity will cost about 50 to $60 billion. Just to give you a sense of what one gigawatt means, it's enough to power 700,000 to 1 million homes, essentially a small Indian city. This is enormous.”

Given this scale, she said, not all AI applications in India would require such capacity.

“But many of the things that we are going to be using AI for don't require this level of capacity and capability. So what are those things? Again, let me bring some visualization to this,” she said.

“These are things like AI services to farmers, citizen services, vernacular services, workflow models for MSMEs, education and learning capacity and capabilities for 200 million students in India.”

Ipsita also reflected on India’s unique development path, referring to the Union Budget presented on February 1.

“The Honorable Finance Minister, when she released the budget on February 1st, talked about India's areas of focus. And while she talked about data centers and AI and technology, on the other side, she talked about 30 million workers in the coconut industry. And she talked about bringing AI to farmers. And that is what describes India,” she said.

She argued that India has consistently adapted technology to local realities. “India has a unique way of taking technology and adopting it to what it needs to be for India, and probably better than anywhere else in the world, as the digital payment stack has shown us.”

Highlighting the country’s digital scale, she added, “If you add up digital payments of the US and China combined in any given day, they don't add up to the digital payments made in India in the same day. I have no doubt that AI will have the same effect on India and through India.”

Concluding her remarks, Ipsita said, “Truly, I think we should both embrace the ambition of hyperscaling with the pragmatism of SLMs on the edge.”