Tech Mahindra has begun 1‑trillion‑parameter sovereign LLM for IndiaAI Mission

Tech Mahindra has said it is developing an indigenous sovereign large language model with one trillion parameters as part of the Government of India’s IndiaAI Mission, a scale the company has described as placing it among the largest AI models currently under development worldwide.

What has been announced?

Chief executive Mohit Joshi has told analysts that Tech Mahindra has been selected under the IndiaAI Mission to help build foundational models and has been working on a sovereign LLM targeting one trillion parameters. The clarification has come alongside the firm’s September‑quarter earnings disclosure.

How this fits into India’s sovereign AI push

In September 2025, the government selected eight additional organisations, including Tech Mahindra, IIT Bombay’s BharatGen consortium and Fractal Analytics to develop foundational models under the IndiaAI Mission. This cohort has followed an earlier wave of selections announced in the first half of 2025.

BharatGen has received 988.6 crore in funding under the Mission and has been tasked with building models that scale up to one trillion parameters, reinforcing the national focus on sovereign AI capabilities across languages and sectors.

The scale: why one trillion parameters matters

Parameter count has served as a rough proxy for representational capacity in modern LLMs; a trillion‑parameter target has indicated both the ambition and complexity of the IndiaAI effort. Officials have also expanded compute capacity: a new tender last month has added about 3,850 GPUs and, for the first time, 1,050 Google Trillium TPUs to the Mission’s national AI cluster.

Background: Tech Mahindra’s LLM work to date

The company has previously launched Project Indus, an Indic language foundation model initiative, and has framed a broader “AI Delivered Right” strategy to scale responsible, production‑grade AI for enterprises.

Has Tech Mahindra built a sovereign LLM before?

Tech Mahindra has developed and open‑sourced Project Indus to support Indic languages, but the one‑trillion‑parameter sovereign LLM now under way has been positioned as part of the government’s IndiaAI Mission, distinct from the earlier in‑house model.

The wider landscape

India’s programme has already seen multiple firms and consortia tapped to develop home‑grown models. Earlier in 2025, Sarvam AI has been selected to build the country’s first sovereign LLM with reasoning and voice capabilities under the Mission, illustrating the multi‑partner approach to foundational model development.

What Tech Mahindra has said

Joshi has said the one‑trillion‑parameter effort “has been a significant technical milestone” aligned to the company’s strategy to “unlock transformation, drive productivity, and accelerate innovation” for customers, and to IndiaAI’s objective to build sovereign AI infrastructure.