TLDR: India’s ambition to become a global leader in Artificial Intelligence is increasingly linked to the adoption and promotion of open-source software. Recent government initiatives, such as the IndiaAI Mission, are channeling significant investments into AI startups that commit to open-source licensing, signaling a strategic shift towards collaborative and accessible AI development.
India’s journey towards establishing itself as a formidable force in Artificial Intelligence is poised to be significantly shaped by the embrace of free and open-source software (FOSS). This perspective challenges the notion that India must solely focus on developing proprietary large language models like GPT-5 or Gemini-X to compete on the global AI stage.
According to Sai Rahul Poruri, CEO of FOSS United, a non-profit dedicated to promoting free and open-source software in India, true open-source software must be freely accessible, modifiable, and redistributable, adhering to the criteria set by the US-based Open Source Initiative (OSI). Poruri highlighted that a ‘research-only’ license, previously used by generative-AI company Sarvam, does not qualify as genuinely open-source. He emphasized that without releasing ‘weights’—the learned parameters that make an AI model intelligent—users are left with a ‘locked box’ that cannot be tinkered with or improved.
In a significant move, the Indian government, under its ambitious Rs 10,000 crore IndiaAI Mission, recently allocated Rs 98 crore in compute subsidies to Sarvam. Crucially, this funding comes with the condition that Sarvam adopts a ‘permissive’ open-source license. This strategic decision places Sarvam, one of eight startups selected to access India’s new compute cluster featuring over 10,000 high-end GPUs like Nvidia H100 and H200, ahead of even some global players like Meta’s Llama, which, despite releasing model weights, maintains commercial-use restrictions. Poruri affirmed that open-sourcing state-funded foundational AI models aligns with FOSS definitions, provided valid copyleft or permissive licenses are used.
Beyond philosophical alignment, there is a strong economic rationale for FOSS in India. A notable example is Kerala’s IT@School project (now KITE), which reportedly saved an estimated Rs 3,000 crore by opting for FOSS over proprietary software. Globally, the open-source AI race is intensifying, with major players like Alibaba Group releasing their Qwen models and startups such as Deepseek launching powerful open-weight models like R1, trained for a mere $6 million. This global trend underscores the growing recognition of open-source as a catalyst for innovation and cost-efficiency in the AI domain.
Also Read:
- Generative AI Propels India’s Next Wave of Application Modernization and Digital Transformation
- India’s AI Imperative: Sarvam AI Co-founder Calls for Accelerated R&D and Investment
FOSS United, co-founded by Zerodha CTO Kailash Nadh and ERPNext founder Rushabh Mehta, continues to advocate for genuinely open software, asserting that this distinction is vital for India’s burgeoning AI ecosystem. The government’s recent policy shift with Sarvam signifies a critical step towards fostering a truly open and collaborative AI landscape in India, leveraging ‘free code’ to build its AI fortune.
