Smaller, Smarter, Stronger: How SLMs Are Fueling India's Grassroots Tech Growth India's linguistic diversity, regional disparities, and mobile-first user base make SLMs particularly compelling

After the buzz around Large Language Models (LLMs), the next big wave in artificial intelligence (AI) is being led by Small Language Models (SLMs). These compact, efficient, and context-aware models are fast becoming a cornerstone of India's digital ambitions—especially for bridging the digital divide and enabling inclusive innovation across Tier-2, Tier-3, and rural regions.

SLMs, unlike their heavyweight counterparts, require significantly less computational power. They typically operate with a few million to a few billion parameters—far less than LLMs like ChatGPT-4, which is estimated to have around 1.8 trillion parameters. Despite their smaller size, these models are proving powerful enough to drive real-world impact, especially in linguistically and culturally diverse markets like India.

According to MarketsandMarkets, the global SLM market stand at USD 0.93 billion in 2025 and is projected to reach USD 5.45 billion by 2032, expanding at a Compound Annual Growth Rate (CAGR) of 28.7 per cent. This surge reflects a growing belief among Indian businesses and policymakers that SLMs are more aligned with the nation's unique digital needs.

Why India needs SLMs

India's linguistic diversity, regional disparities, and mobile-first user base make SLMs particularly compelling. S Anjani Kumar, Partner at Deloitte India explains, "Developing a few specialised small language models over a single general-purpose large language model is better suited because the problem statements in India are diverse and unique. Over time, organisations will build a model garden and could deploy bespoke models for specific use—for example, an SLM for the finance function in an insurance company."

Neeti Sharma, CEO of TeamLease Digital, echoes the sentiment by emphasising infrastructure advantages, "SLMs are cheaper to build and run. They don't need big servers or fast internet—they can work on mobile phones and basic devices. This makes them perfect for villages and small towns where internet and electricity can be a problem. They also save energy and keep data safe by running on local systems."

As per the PIB, 95.15 per cent of Indian villages have 3G/4G internet access as of April 2024, making low-resource AI models like SLMs practical for rural deployment.

India's mobile-centric market further strengthens the case for SLMs. According to Statcounter (April 2025), mobile phones account for 79.49 per cent of web traffic in India, compared to just 19.9 per cent from desktops.

The real-world impact

In key sectors such as governance, healthcare, education, and banking, SLMs are beginning to demonstrate measurable impact. Priyanka Kulkarni, Manager – Telecom, Media and Technology at Aranca says, "SLMs support local data processing, aligning with India's data sovereignty and privacy goals. They lower the barrier to entry for AI innovation. Startups, research labs, and even state governments can build and iterate AI models without massive datasets or supercomputing resources."

Referring to recent independent findings, Kulkarni notes that vertical-specific SLMs might deliver tangible results. For instance, in the BFSI sector, companies could achieve up to 70 per cent cost reduction in contact centres and a 75 per cent decrease in delinquency rates through vernacular AI adoption.

Adding to this, Neeti Sharma says, "SLMs are helping banks approve loans faster, aiding AIIMS with local-language medical advice, and supporting tribal students through regionally tailored content. They're transforming access and equity across sectors."

Building trust and inclusion

Beyond performance, SLMs are advancing ethical AI principles by ensuring inclusivity and local relevance. Ankush Sabharwal, Founder and CEO of CoRover, which is building BharatGPT Lite in 14 Indian languages says,"We ensure accuracy by training our SLMs on rich, multilingual datasets. Bias mitigation is achieved through balanced datasets representing all regions and communities. Local relevance is maintained with continuous refinement based on user feedback."

This approach has enabled virtual assistants developed by CoRover to assist institutions such as IRCTC, LIC, MaxLife, and local police departments, improving citizen interaction and support in regional languages.

Economic empowerment through AI

SLMs hold the potential to unlock opportunities for Bharat's next 500 million users, many of whom remain on the fringes of the digital economy. Mahesh Kumar, CPTO and Co-founder of Gigin AI elaborates, "SLMs transform technology from daunting to user-friendly by enabling AI that understands regional contexts, speaks local languages, and operates on affordable devices. They enable students to learn in their mother tongue, workers to search for jobs through voice commands, and farmers to get agriculture advice in their dialect."

Such grassroots-level access to technology is key to democratising economic participation and reducing urban-rural disparities.