India’s IndiaAI Mission recently announced that it has selected three more AI startups—Soket AI, Gnani.ai, and Gan.AI—for developing indigenous foundation models. With this, a total of four startups are now working under this initiative including Sarvam AI.

Why does India need foundation models?

Foundation models are large-scale pre-trained AI systems (such as OpenAI’s GPT, Google’s Gemini). These models are trained on vast amounts of data and can be adapted for a wide range of tasks. Developing India’s own foundation models would enable better solutions for Indian languages, culture, and local problems. Currently, most major models are developed in Western countries.

With the development of these foundation models, high-performance GPUs can be provided to Indian scientists and startups, accelerating their research. Additionally, Indians are currently reliant on foreign AI platforms for using AI technologies. Through the IndiaAI Mission, India aims to reduce its dependence on foreign AI platforms and become self-reliant in the field.

IndiaAI Mission: The Indian Government’s Vision

The Indian government aims for its AI startups to rank among the top in the global AI landscape. With this goal in mind, Union IT Minister Ashwini Vaishnaw stated, “These three teams also have a very big target ahead of them. Whatever area they focus on, they must be among the top five in the world. This is a clear goal.”

Soket AI, founded by Abhishek Apporwal, will build India’s first open-source 120-billion parameter foundation model trained on two trillion tokens. It will be optimized for the country’s linguistic diversity and designed to target sectors such as defense, healthcare, and education.

Similarly, Gnani.ai, co-founded by Ganesh Gopalan and Ananth Nagaraj, will develop a 14-billion parameter voice AI foundation model that delivers multilingual, real-time speech processing with advanced reasoning capabilities. This will make generative AI more accessible to the country’s non-English speaking population. Gan.ai aims to create lightweight and domain-specific language models. It focuses on foundational models for audio and video and is developing tools for AI avatar video generation.

Vaishnaw further shared that one of the selected startups has expressed interest in becoming number one or two in their chosen domain. This has significantly boosted the enthusiasm of the other startups. “This is the mindset with which the teams are working. I’m happy,” he said. He also noted that this initiative could help bring Indian talent back from abroad, potentially reversing what was once a brain drain into a “reverse brain drain.”

Progress of the IndiaAI Mission

Vaishnaw shared that 367 datasets have already been uploaded to AI Kosh, India’s curated open repository of AI datasets.

Alongside the selection of new startups, the government has announced the second round of AI compute panelment under the IndiaAI Compute Pillar. This includes an additional 15,000 GPUs on top of the existing 18,000, bringing the total available GPU compute capacity to 34,000.

The newly listed companies for GPU sourcing are Netmagic (NTT Global), Cyfuture India, Sify Digital Services, Vensysco Technologies, Locuz Enterprise Solutions, Yotta Data Services, and Ishan Infotech.

Cyfuture India will supply a variety of GPUs, including NVIDIA H100, L40S, A100, AMD MI300X, MI325X, and Intel’s Gaudi 2 and Gaudi 3. Ishan Infotech will offer the H100, H200, and L4 models.

Netmagic will supply H100, H200, L40S, and L4 GPUs along with AMD MI300X. Sify Digital Services will provide H100, H200, and L4. Locuz Enterprise Solutions will supply H200. Vensysco Technologies will offer H100 and A100. Yotta Data Services will supply NVIDIA B200.

This enhanced computing capacity in the cloud will offer a unified platform for sourcing critical GPUs essential for the core development of AI solutions.

So far, the IndiaAI Mission’s compute pillar has been operated through public-private partnerships with companies like Jio Platforms, Nxtra Data Center, Locuz Enterprise, E2E Networks, CtrlS Data Centers, CMS Computers, Orient Technologies, Tata Communications, Vensysco, and Yotta Data Services.

Conclusion

Indian companies have managed to carve out a space in the global AI landscape. However, to sustain their position in this competitive field, these companies must further strengthen their AI models while also ensuring they can deliver high-quality AI solutions at a relatively lower cost.

There are, however, some challenges—particularly the need for high-quality computing infrastructure, large and reliable datasets, and a shortage of deep AI researchers. Still, due to investments, international collaboration, and rapid innovation, Indian companies are advancing quickly in the global AI race and are expected to become even stronger competitors in the coming years.