India Selects Three More Startups to Develop Indigenous AI Foundation Models

With 34000 GPUs, IndiaAI Mission Backs Soket AI, Gnani.ai, Gan.AI for Foundation Models | AIM

Ashwini Vaishnaw shared that 367 datasets have already been uploaded to AI Kosha, India's open repository of curated AI datasets. IndiaAI Mission on Friday announced the selection of three more startups, Soket AI, Gnani.ai, and Gan.AI, to develop indigenous foundation models. This brings the number of startups under the foundation model development initiative to four, including the recently announced Sarvam AI. "These three teams also have a very big target in front of them. Whichever sector they focus on, they must become among the top five in the world. That's a clear target," said Union IT Minister Ashwini Vaishnaw. Soket AI founded by Abhishek Upperwal, will develop India's first open-source 120 billion parameter foundation model optimised for the country's linguistic diversity, targeting sectors such as defence, healthcare, and education. Similarly, Gnani.ai, co-founded by Ganesh Gopalan and Ananth Nagaraj, will build a 14 billion-parameter Voice AI foundation model that delivers multilingual, real-time speech processing with advanced reasoning capabilities. Gan.ai focuses on foundational models for audio and video and is building AI avatar video creation tools. Vaishnaw added that one of the selected startups expressed interest in becoming number one or two in its chosen domain. "That's the spirit with which teams are working. I'm glad," he said. He also noted that the initiative could help bring back Indian talent from abroad, and what used to be a brain drain should now be a reverse brain drain. Vaishnaw also shared that 367 datasets have already been uploaded to AI Kosha, India's open repository of curated AI datasets. Alongside the new startup selections, the government announced the second round of AI compute empanelment 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 new empanelled 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 provide a range of GPUs, including NVIDIA H100, L40S, A100, AMD MI300X, MI325X, and Intel's Gaudi 2 and Gaudi 3. Ishan Infotech will offer 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 offer the H200. Vensysco Technologies will provide H100 and A100. Yotta Data Services will supply the NVIDIA B200. This expanded compute capacity on cloud will provide a common platform for sourcing of critical GPUs for key tasks in the development of AI solution. The IndiaAI mission's compute pillar has so far operated through public-private partnerships with companies such as Jio Platforms, NxtGen Data Centre, Locuz Enterprise, E2E Networks, CtrlS DataCenters, CMS Computers, Orient Technologies, Tata Communications, Vensysco, and Yotta Data Services.

After Sarvam.ai, Soket AI, Gnani.ai, Gan.ai to now build foundation AI models, says IndiaAI Mission

The IndiaAI Compute portal now has 34,000 GPUs, sourced from Nvidia, AMD, AWS, and Intel. In the second round of the GPU tender, seven technically qualified bidders have been empanelled. IndiaAI Mission has now selected Soket AI, Gnani.ai, and Gan.ai, in addition to Sarvam.ai, to develop foundation AI models.The IndiaAI Mission on Friday announced Soket AI, Gnani.ai, and Gan.ai will be developing foundation artificial intelligence (AI) models apart from Sarvam.ai, which was initially selected. ET had reported this more than a month ago. Soket AI will develop an open source 120 billion parameter foundation model optimised for the country's linguistic diversity targetting sectors such as defence, healthcare and education. Gnani.ai will build a 14 billion parameter voice AI foundation model delivering multilingual, real-time speech processing with advanced reasoning capabilities. Gan.ai will create a 70 billion parameter multilingual foundation model targeting superhuman text-to-speech (TTS) model capabilities. The IndiaAI Mission also said that 34,333 GPUs are now available on the IndiaAI Compute portal. These include GPUs from Nvidia, AMD, AWS, and Intel. The offering comprises 15,100 H100 Nvidia GPUs, 8,192 Nvidia B200 GPUs, 4,812 Nvidia H200 GPUs, and 1,973 Nvidia L40S GPUs. The other GPU variants each comprise fewer than 1,000 or 500 in number. All seven technically qualified bidders in the second round of the GPU tender are empanelled, it said. The seven empanelled bidders are Cyfuture India, Ishan Infotech, Locuz Enterprise Solutions, Netmagic IT Services (NTT GDC India), Sify Digital Services, Vensysco Technologies, and Yotta Data Services. 367 datasets have been uploaded on AI Kosh, Vaishnaw revealed. AI Kosh is the government's platform that provides a repository of datasets, models and use cases to enable AI innovation. It also features AI sandbox capabilities. IndiaAI Mission on May 16 had said that it has received 506 foundation AI proposals across the three phases. Given the overwhelming response and continued interest, the mission has decided to extend the deadline for submissions under phase 3 of the call for proposals. The earlier deadline was April 30. "Further dates for submission of proposals, the acceptance of new applications post April 30, will be announced subsequently, as per requirements, once the examination of proposals already submitted has been completed," IndiaAI said on its website. As part of the first phase of approvals, Sarvam.ai was selected to initiate the development of an indigenous foundational model. Sarvam's multimodal, multi-scale model will have 70 billion parameters. Sarvam cofounder Vivek Raghavan had said on April 26 that the model will be completed in six months. Capable of reasoning, designed for voice, and fluent in Indian languages, Sarvam's model will be ready for secure, population-scale deployment, the startup had said in a statement. According to cofounder Pratyush Kumar, Sarvam is developing three model variants -- Sarvam-Large for advanced reasoning and generation, Sarvam-Small for real-time interactive applications, and Sarvam-Edge for compact on-device tasks. Sarvam will get access to 4,096 Nvidia H100 graphics processing units (GPUs) for six months from the IndiaAI Mission's common compute cluster to train its model, people in the know had said. "We are collaborating with AI4Bharat at IIT-Madras, a leader in Indian language AI research, to build these models," Kumar had said.

Soket AI, Gnani.ai & Gan.ai To Develop Indigenous AI Models

The government received 506 applications till April 30 since the launch of the call for proposals under the IndiaAI Mission in January this year After SarvamAI, the IT ministry has now selected three more AI startups - Soket AI Labs, Gnani.ai and Gan.ai - to build large language models (LLMs) under the IndiaAI mission. In a statement, the IT ministry said that Soket AI Labs will develop India's first open source 120 Bn parameter foundational model optimised for the country's linguistic diversity. The model will target use cases in sectors such as defence, healthcare and education. Meanwhile, Gnani.AI will build a 14 Bn parameter Voice AI foundational model to offer multilingual and real-time speech processing with advanced reasoning capabilities. Further, Gan.ai will create a 70 Bn parameter multilingual foundational model, which will boast "superhuman TTS" (text-to-speech) capabilities at par with current global leaders. The project will be undertaken under the aegis of the foundational model pillar under the IndiaAI Mission, which aims to develop and deploy indigenous functional models trained on India-specific data. The government received 506 applications till April 30 this year since the launch of the call for proposals under the IndiaAI Mission in January this year, the IT ministry said. "We're proud to lead the way in developing voice-to-voice large language models for India and the world, because we believe transformative AI must speak the language of the people it serves," said Ganesh Gopalan, cofounder and CEO of Gnani.ai. Founded in 2017 by Gopalan and Ananth Nagaraj, Gnani.ai currently has speech engines (software that converts written text into spoken audio) for 10 Indian languages. Its client base includes TVS Credit, Bajaj Group, Fibe and Muthoot Finance. The startup has filed over 12 patents to date. Meanwhile, Soket AI Labs, incorporated in 2019 by Abhishek Upperwal, claims to have built the country's first open-source multilingual model on top of TinyLlama, which leverages Meta's Llama 2 architecture. Part of NVIDIA's Inception Programme and AWS Activate, the startup claims to be building full-stack AI solutions for enterprises. Gan.ai, founded in 2021 by Suvrat Bhooshan, claims to specialise in creating and using AI-powered avatars for video communication. It leverages generative adversarial networks (GANs) to generate realistic and lifelike avatars that can deliver personalised messages. This comes a month after the Centre shortlisted Bengaluru-based GenAI startup Sarvam AI to helm the project to build India's first homegrown sovereign LLM. SarvamAI has reportedly received a dedicated compute infrastructure to build the model from scratch. The model will focus on reasoning, support voice-based tasks and work fluently across Indian languages. It is being designed for secure, population-scale use. Notably, the LLM will be fully developed, deployed and optimised within the country, using local talent and infrastructure. It is expected that the three newly selected startups will also receive similar sops and incentives to create their respective LLMs. India's AI Mission Takes Shape IT minister Ashwini Vaishnaw yesterday urged the newly selected teams to aim for "top-five global positions" in their respective sectors. "Technology should not be left in the hands of a few. It's very important that a larger section of the society should be able to access technology, develop new solutions and get better opportunities. That's the philosophy with which IndiaAI Mission was created. We are making significant progress in practically every pillar of the AI mission. Common compute is a very important principle of democratizing technology," said Vaishnaw. In a statement, the IT ministry said India's national compute capacity has crossed the 34,000 graphics processing unit (GPU) mark. This comes a few days after Vaishnaw reportedly said that the government was all set to procure an additional 14,000 GPUs under the IndiaAI mission. This was in addition to the 18,000 GPUs already procured by the Centre in the past few months. Meanwhile, Vaishnaw also noted that 367 datasets have already been uploaded on the AI Kosh portal. For context, the platform offers access to datasets, AI models, and tools to accelerate AI innovation in the country. The minister reiterated that IndiaAI Mission will look to create a comprehensive ecosystem, encompassing foundational models, compute capacity, safety standards, and talent development initiatives, to make the country an AI hub. The Union Cabinet approved the IndiaAI Mission in March 2024, with an outlay of INR 10,372 Cr spanning a period of five years. It comprises initiatives such as IndiaAI Compute Capacity, IndiaAI Innovation Centre (IAIC), IndiaAI Datasets Platform, IndiaAI Application Development Initiative, IndiaAI FutureSkills, IndiaAI Startup Financing, and Safe & Trusted AI.