Abstract

With the United States and China dominating the race for AI, and building capabilities across the full stack of AI, India needs to take radical action to defend its strategic autonomy. In light of the massive lead that the U.S. and China have in AI, the fastest, most cost-effective, and pragmatic path forward for India is to build a global AI commons through open source collaborations. The essay proposes measures spanning the IndiaAI Mission, academic research, open-source communities, international collaboration with the Global South and other partners, participation in global standards-setting, and public procurement. It argues that sustained investment in open-source AI can strengthen India’s technological autonomy while enabling more inclusive global innovation.

Background

On June 12, 2026, Anthropic, the leading U.S. AI company received an export control directive from the U.S. Government to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. In response, Anthropic abruptly disabled access to Fable 5 and Mythos 5 for all customers to ensure compliance. In a statement, Anthropic said that the directive from the U.S. Government did not provide specific details of its national security concern.

This development came a year after India was hit with 50 per cent tariffs on exports to the U.S., and severe limitations on imports of AI hardware. The tariffs were struck down by the U.S. Supreme Court, but these developments illustrate that we are entering an era of geopolitical uncertainties and political turbulence that have huge implications for India’s digital sovereignty. Therefore, it is instructive to look back and examine how India overcame similar situations in the past. 

Wheat and nuclear power were two major flashpoints in the India-U.S. relationship. In 1965-66, India faced severe famine and needed to import wheat from the U.S. The government, under President Lyndon B Johnson, leveraged this situation to pressure India, which had criticised America’s bombings during the Vietnam War. Johnson made wheat exports to India conditional upon assurances that the Indian government would implement agricultural reforms and temper its criticism of America’s Vietnam policy.

When the Government of India (GOI) protested and pointed out that its position was not different from what the United Nations and the Pope had said, Johnson reportedly said, “The Pope and the [UN] Secretary-General do not need our wheat.”

President Johnson’s “ship-to-mouth” tactics humiliated India and became the catalyst for the Green Revolution, which ensured that India would never have to depend on another country for essential foodgrains. The government created buffer stocks through the Food Corporation of India (FCI), invested in building dams and canals, fertiliser plants, imported High Yielding Variety (HYV) seeds from countries like Mexico, and promoted agricultural research to ensure that the country never again faced such a dire situation. 

Similarly, in 1974, when India conducted its first nuclear test, several countries, including the U.S. and Canada, imposed sanctions that cut off access to nuclear technology and enriched uranium. The Tarapur Atomic Power Station (TAPS), which was built under a 1963 Indo–U.S. agreement, was particularly affected as the U.S. suspended supplies of reactor parts, processing equipment, and uranium. Canada and other U.S. allies also curtailed cooperation, while the subsequent formation of the Nuclear Supply Group (NSG) pushed India into a period of nuclear isolation that lasted 30 years. 

In response, India accelerated its Pressurised Heavy Water Reactors (PHWR) programme and shifted away from dependence on U.S.-supplied enriched uranium towards natural uranium and heavy water reactors. Uranium mining was expanded, and Heavy Water Production Plants were set up at Kota, Baroda, and Tuticorin, to ensure self-sufficiency. In 1998, when India conducted another series of nuclear tests, it faced a fresh round of sanctions. Despite these challenges, India continued to invest in an indigenous nuclear power program designed to leverage its vast reserves of thorium. 

This decades-long effort bore fruit on 6 April 2026, when the indigenously designed and built Prototype Fast Breeder Reactor (PFBR) at Kalpakkam attained first criticality. Once fully operational, the PFBR will make India the second country after Russia to operate a commercial-grade fast breeder reactor. This marks a major step towards long-term energy independence, particularly as India aims to increase its nuclear power production from 8.78 Gigawatt (GW) to 22.38 GW by 2031–32. At a time when the blockade of the Strait of Hormuz has sharply increased India’s oil import costs, exploring multiple sources of power generation becomes a critical imperative. 

This paper follows Pohle and Thiel in conceptualising sovereignty as control over critical infrastructure, economic freedom, and individual agency. It is also underpinned by Floridi’s approach of seeing power as control. Building on these perspectives, the paper argues that a sovereign state needs to control and protect its AI infrastructures against state and non-state actors, as well as from domestic and foreign threats, through legitimate means available to it. 

The Global and Indian AI LandscapeWhen it comes to AI capabilities, countries can be divided into three tiers: 

• Tier 1: Countries with full-stack capabilities across hardware, software, cloud computing, and AI skill sets. At present, only the U.S and China (with the notable exception of high-end AI processors) meet these criteria.  
• Tier 2: Countries with partial capabilities across the stack. Countries in the European Union, India and a range of countries across Asia, Africa and South America meet these criteria. 
• Tier 3: Countries with very limited capabilities. This includes, for example, Pacific Ocean countries that have very small populations and limited resources.  

The United States leads global AI investment in 2025, with approximately USD 471 billion, followed by China at around USD 119 billion and India at about USD 11.1 billion. These figures, which include both public and private sector funding, highlight the vast disparity in AI investment among the world’s biggest economies. Four American companies – Meta, Amazon, Alphabet, and Microsoft – alone are set to spend a record USD 320 billion on AI and data centres in 2025, up from USD 230 billion in 2024. 

Source: Spherical Insights

The European Union has also announced plans to mobilise USD 200 billion (about USD 212 billion) in AI investments between 2025 and 2027 through its InvestAI initiative. The programme aims to support the open and collaborative development of the most complex AI models while positioning Europe as a global AI hub. The initiative blends public and private funding to build advanced AI infrastructure, support research, and deploy large-scale AI factories and gigafactories across member states. More broadly, it aims to strengthen Europe’s technological sovereignty, accelerate the adoption of AI across the EU, and triple the region’s AI compute capacity.

To respond to the rapid growth of AI, GOI sanctioned INR 10,300 crore (about USD 1.2 billion) for the IndiaAI Mission in 2024. The mission aims to “catalyse various components of the IndiaAI Mission, including pivotal initiatives like the IndiaAI Compute Capacity, IndiaAI Innovation Centre (IAIC), IndiaAI Datasets Platform, IndiaAI Application Development Initiative, IndiaAI FutureSkills, IndiaAI Startup Financing, and Safe & Trusted AI.” 

The key initiatives under the IndiaAI Mission have been:

• IndiaAI Compute Capacity: Aims to create a high-end AI computing ecosystem with access to over 38,000 GPUs to support AI development. These GPUs have been made available at a rate of INR 65 per hour (USD 0.72). 
• IndiaAI Innovation Centre: Focuses on developing indigenous Large Multimodal Models (LMMs) and foundational models. 
• IndiaAI Datasets Platform: Aims to provide unified access to quality non-personal datasets for research and startups. 
• Centres of Excellence (CoEs): Established in sectors like healthcare, agriculture, and sustainable cities, with a new CoE for education announced in the 2025 Union Budget. 
• Bharat Data Exchange: An extension of the government’s open data platform, creating a centralised data exchange. 

Even as India builds its AI capabilities through the AI Mission, global players have been making aggressive moves to expand their presence in the Indian market through partnerships with Indian telcos. Perplexity, for example, signed an agreement to offer free access to Perplexity Pro, priced at USD 200 annually, to Airtel’s 360 million subscribers. Google soon followed suit by offering free access for 18 months to its Gemini AI Pro plan priced at USD 396 to Reliance Jio’s 500 million subscribers. Not to be left behind, OpenAI announced that it would make ChatGPT Go available free of charge for one year to Indian users.

While these offers help increase the monthly active user base of these US-based platforms and help them partly justify their sky-high valuations, building upon and using these services introduces a new set of risks:

1. Denial of Technology: The Microsoft-Nayara case illustrates how access to critical digital infrastructure can be disrupted when services are hosted and controlled outside a country’s jurisdiction. In July 2022, Microsoft disconnected cloud services to Indian refining and marketing company Nayara Energy without any explanation. Microsoft stated that it did so to comply with the European Union’s (EU) sanctions against Russian-origin businesses supplying goods and services to certain geographies. 
2. Vendor Dependency: Building any AI system on top of a closed source, cloud-based AI system creates the risk of vendor lock-in and long-term dependence on the vendor for feature upgrades and security patches. In some cases, as with Builder.AI, the vendor could go out of business. As CTO Magazine notes, “AI vendor lock-in occurs when your organisation becomes so reliant on a single AI or cloud provider that detaching from it becomes technically, financially, or legally prohibitive.” 
3. Surveillance: Foreign-hosted AI services may also create surveillance and data-governance concerns. As the American Civil Liberties Union notes, the National Security Agency identifies non-U.S. persons it wishes to monitor, and orders U.S. technology companies to disclose all communications and data to and from those accounts. Section 702 of the Foreign Intelligence Surveillance Act permits U.S. intelligence agencies to surveil the communications of non-U.S. nationals without a warrant. If the USA’s surveillance list includes Indian diplomats, military personnel, politicians and bureaucrats, U.S.-based AI platforms will have no option but to hand over their data to the authorities. Similar concerns arise in relation to AI services hosted in China, given the tight control that the Chinese government maintains over Chinese technology firms. These concerns illustrate that dependence on AI services from Tier 1 countries is fraught with surveillance risk. 

Lessons from OSS Implementations in India

Given the fragility of contemporary AI supply chains and the possibility of disruption, neither of the Tier 1 countries can be relied upon for a guaranteed supply of AI hardware and cloud computing solutions. However, Open Source AI (OSAI) models – which can be examined, verified, freely built upon without restrictions, and hosted domestically – offer a viable alternative, even if they originate from Tier 1 countries. Both the Indian government and private sector have extensive experience in leveraging open source software (OSS), and these experiences offer valuable lessons for the adoption of OSAI.

For example, India’s Ministry of Electronics and Information Technology adopted a Policy on Adoption of Open Source Software for Government of India in 2015. This policy aims:

• To provide a policy framework for rapid and effective adoption of OSS
• To ensure strategic control over e-governance applications and systems from a long-term perspective.
• To reduce the total cost of ownership (TCO) of projects.

The policy reads as follows:

 “Government of India shall endeavour to adopt Open Source Software in all e-Governance systems implemented by various Government organizations, as a preferred option in comparison to Closed Source Software (CSS). 

The Open Source Software shall have the following characteristics:

3.1. The source code shall be available for the community / adopter / end-user to study and modify the software and to redistribute copies of either the original or modified software.

3.2. Source code shall be free from any royalty.”

GOI could adopt a similar approach by mandating the use of OSAI in critical e-governance projects while modifying the policy to reflect the unique needs of AI. 

Some of the most notable deployments of OSS in e-government projects are Aadhaar, the world’s largest biometric ID system; CoWIN, the COVID-19 vaccination platform; DigiLocker, the GOI’s digital document storage platform; DIKSHA, the national platform for school education; Aarogya Setu, the COVID-19 contact tracing app, and the Goods and Services Tax Network. Together, these initiatives demonstrate the government's ability to deploy OSS at a national scale.

The Indian private sector has also been an active adopter of OSS. According to a report from the National Law School of India University (NLSIU), The Rise of FOSS in India: Empirical Evidence and Insights from Cross-Sectoral Case Studies, organisations such as Zerodha, Remiges, and the National Payments Corporation of India have successfully incorporated Free and Open Source Software (FOSS) into their operations. The report analyses the adoption of FOSS in India, primarily through case studies across four sectors (healthcare, education, finance, and software and IT services) and different types of organisations (start-ups, non-profits, medium, large, and public sector organisations), and highlights both the benefits and challenges experienced by organisations using FOSS. Its findings suggest that while organisations benefit from increased innovations, cost/ time savings, flexibility, and enhanced security, they also face challenges such as a shortage of skilled personnel and limited community support. Organisations are also found to take a cautious approach to licensing, favouring permissive licenses over restrictive ones. Based on these diverse empirical findings, the report recommends a range of policy reforms, including mandating FOSS adoption for government bodies, integrating FOSS in Information and Communication Technology (ICT) education, and updating procurement guidelines to mandate FOSS solutions when they are available.

Taken together, these examples demonstrate that both the government sector as well as the private sector have proven capabilities in implementing large-scale OSS projects, and these lessons can be relevant to OSAI implementations. 

The Way Forward

A distinctive advantage of open models is that it enables India to be part of global value chains while retaining its digital autonomy. If we build on top of closed, cloud-based systems like OpenAI, we are dependent on an organisation that is located in another geography. This exposes us to the denial of technology, vendor dependency and surveillance risks mentioned earlier. If these risks materialise, substantial investments of time, capital, and human resources dedicated to building applications on those platforms could be undermined.

GOI has been cognisant of these challenges and has ensured that projects of national importance are built on top of OSS, with support from Indian software vendors. These OSS deployments have been extremely successful in helping India save money and protect the country’s digital sovereignty. At the same time, we have to recognise that these are primarily inward-looking efforts and not aligned with India’s ambitions of building a global AI Commons. Speaking at the Global AI Summit held in New Delhi in February 2026, Indian Prime Minister Narendra Modi said, “When codes are open and shared, millions of young minds can make them safer and better. Some countries and companies believe AI is a strategic asset and must be developed confidentially. But India thinks differently. We believe AI will benefit the world only when it is shared.”

To truly realise the vision of AI as a global common good, India needs a fundamental shift in its approach to open source. India has done a great job of consuming OSS and deploying it for domestic needs. Now it needs to switch gears and become a contributor and leader of the global OSS community. The IndiaAI Impact Summit, which was attended by world leaders and global AI industry leaders, shows that India has the convening power. It needs to build on this by working with like-minded global institutions and governments, the international open source community, and the Global South to create an open source roadmap for achieving AI for All. 

At this juncture, it is appropriate to ask the counterfactual question: Can the goal of AI for All be achieved through Closed Source Software (CSS)? 

CSS licenses give users limited rights to use the software. Unlike OSS, the rights to review, modify, and share modifications are not granted by CSS licenses. The source code of CSS is protected through trade secrets and only the original creators have the right to modify the code. CSS, therefore, does not serve as a platform for collaboration and cannot be freely shared. 

In contrast, OSS is driven by the Collaborative Innovation Model (CIM). The three pillars of CIM are community, collaboration, and the shared ownership of knowledge. The Linux Operating System, which now powers everything from servers to smartphones and supercomputers, is an example of CIM. Linux began as a student project of Linus Torvalds, who released it in September 1991. A year later, he changed the license from a non-commercial license to the GNU General Public License, which gave users the right to review, modify, and share those modifications with others. This drove unprecedented collaboration from users around the world with the Linux code base growing from 10,000 lines in 1991 to 40 million lines of code in 2025. 

Linux is now the world’s largest collaborative software project. More than 12,000 individuals have contributed to the Linux kernel and thousands more have contributed to subsystems like the Graphical User Interface and other software that make up the complete operating system. Thanks to the CIM that powers Linux, the pace of innovation outstripped that of proprietary operating systems like Unix and Windows. 

Speaking about the growth of Linux, Jonathan Corbet, a long-time Linux kernel contributor and editor of Linus Weekly News (LWN), said:

“Unlike the centralized, exclusive development of other Unix-like systems, now largely forgotten, Linux welcomed contributions from anyone willing to submit code. This openness, combined with the GNU General Public License (GPL) 2.0, fostered a unified codebase and prevented fragmentation. Anybody could be a part of it, and as a result, Linux developed a community that you didn't find in other open projects at that time." 

Linux now runs all the top 500 supercomputers in the world, is expected to be the fastest growing server operating system from 2025 to 2034, and powers 67.46 per cent of all smartphones through the Linux-based AndroidOS. 

Linux is one of the most prominent examples of a global, collaborative OSS project, but it is not the only one. GitHub, the world’s largest OSS development platform, reports that it now hosts 630 million OSS projects, 180 million developers, 4.3 million AI projects, and has accepted 1.2 billion contributions to these OSS projects. The Linux Foundation, which started life as a host for the Linux project, now hosts 1,300 open source projects with 89 million lines of code added weekly, and works on 100+ standards and specifications. The Apache Software Project hosts 290 projects, and the Eclipse Software Foundation hosts another 400 projects. 

A study by researchers at the Harvard Business School, Massachusetts Institute of Technology (MIT), The Linux Foundation, and the University of Toronto estimated that the supply-side value of widely used OSS is USD 4.15 billion, while the demand-side value is much larger at USD 8.8 trillion. The researchers found that firms would need to spend 3.5 times more on software than they currently do if OSS did not exist. The supply-side value was estimated by calculating the cost to recreate the most widely used OSS once, while the demand-side value was estimated based on the replacement cost that would be incurred by firms if they were forced to develop these capabilities internally.

Therefore, if India is to realise Prime Minister Modi’s vision of AI that benefits all, OSS is the way forward. However, it must significantly expand its engagement with, and contribution to, the global OSS ecosystem to achieve this goal. 

Building on India’s Strengths

The GitHub report cited earlier reports that India alone added more than 5 million developers this year (over 14 per cent of all new accounts) and is on track to account for one in every three new developers on GitHub by 2030, with 57.5 million developers. India has the talent necessary to make the most of Open Source AI (OSAI) models, and its domestic IT industry has the skills needed to implement both large and small OSAI projects. India has also led the creation of Digital Public Infrastructure like the Modular Open Source Identity Project (MOSIP) that enables countries to build their own ID systems; the Beckn Protocol that enables the creation of decentralised e-commerce networks; the Sunbird open source platform that was used to build DIKSHA (Digital Infrastructure for Knowledge Sharing), a national platform for school education; and many others. 

Note: When we use the term OSAI Models, we specifically refer to those models released under a license approved by the Open Source Initiative (OSI). Many of the self declared “open source models” do not meet the Open Source Initiative’s definition of Open Source AI (OSAI) models. 

India also has the vision, articulated at the highest political level, of an AI ecosystem that is open and shared. However, it lacks a programmatic approach for transforming this vision into reality. To move beyond OSS as a rhetorical device to OSS as a lived reality, India must embrace the ethos of Collaborative Innovation. It needs to move from passively consuming OSS to actively shaping global OSAI projects. The IndiaAI Mission could take the lead in convening the global OSAI community, strengthening existing OSAI projects, identifying gaps in the OSAI ecosystem, and working to address them. Priority areas for OSAI development could include building OSS alternatives to NVIDIA’s proprietary CUDA parallel computing platform and application programming interface and building OSAI hardware using open source RISC-V architecture.

At this point in time, it may seem that closed source LLMs like OpenAI, Gemini and Perplexity have overwhelming momentum and billions of dollars of capital behind them. However, history suggests that capital alone does not ensure success. As the example of Linux demonstrates, the Collaborative Innovation ecosystem ultimately overtook other closed source operating systems like Windows NT and Unix, despite the billions invested in their development and marketing. In the last two years, some OSAI systems have attained parity with proprietary models. Over the next few years, the pace of innovation in OSAI could help it outstrip the capabilities of proprietary AI models. In the next section, we explore how India can leverage OSAI to climb from Tier 2 to Tier 1 status. 

Climbing from Tier 2 to Tier 1

GOI has been a major consumer of OSS as seen in projects like Aadhaar, UPI, GSTN and others. However, as Prime Minister Modi has said, if we have to “develop [emphasis ours] AI as a Global Common Good,” a consumption-based approach will no longer be sufficient. 

In terms of open source maturity for an ecosystem, there are three levels:

• Consuming open source software.
• Contributing back to the open source projects you consume.
• Producing your own open source software.

If India wants to be a Tier 1 AI country and acquire capabilities across the AI stack, it must contribute to the global AI commons and produce/develop global OSAI projects from India. Open source communities measure their health in terms of downloads of the software, number of active contributors to the projects, number of code submissions, and activity on the discussion groups. Attaching a license approved by the Open Source Initiative is only the first step in the open source journey. Without an active open source community, a project is open source in letter but not in spirit. This is the critical gap the GOI needs to address. 

For a start, the GOI could mandate that projects funded through the IndiaAI mission should not just release their systems under OSAI licenses but also build vibrant open source communities around them. The IndiaAI Mission should fund community building, and measures of community health, such as the number of downloads, the number of contributions from outside the funded organisation, and the number of pull requests, should be tracked as Key Performance Indicators (KPIs). 

India still does not have an indigenously created OSAI project that is globally adopted and used. The IndiaAI Mission should therefore prioritise the creation of globally active, inclusive, and scalable OSAI communities rather than focusing solely on releasing projects under open source licenses. 

Building OSS communities is an act that requires sustained planning and effort. Accordingly, the IndiaAI Mission must ensure that OSAI projects funded by it include dedicated budgets for community engagement and measure the health of the open source community that is being built. In the absence of such active community building and maintenance efforts, open source projects often stagnate and eventually die. In the report titled, Building Communities Around Digital Public Goods, the authors say: 

“The typical DPG lifecycle involves creating, deploying, sustaining, and evolving the goods, and communities are critical in the last two stages of the development of DPGs. Communities help the DPG to sustain and evolve by enabling characteristics that propel the adoption of the DPGs. These characteristics include relevancy, modularity, and customisability, among others. Given that the end goal of every DPG creator is to enhance adoption, a community helps bridge the gap between initial adoption and large-scale adoption by helping the DPG evolve and sustain.” 

Private conversations with senior policymakers reveal that GOI’s programmers are worried about encountering criticism when they open up their code to external scrutiny. Consequently, a mindset change is required whereby GOI’s programmers (and its contractors) actively engage with global open source communities, submit bug reports and code patches, and help maintain the software. 

The next stage of maturity is for the GOI to actively lead the creation of global OSS and OSAI projects. Doubling down on the IndiaAI Mission’s existing projects and investing in creating OSS communities around them would be a good starting point. Metrics such as global adoption, active contributors, and long-term community participation could be tracked across all Indian OSAI projects. 

Against the backdrop of capital deployment of hundreds of billions of dollars invested in AI, India’s only viable pathway from Tier 2 status to Tier 1 status lies in open source. India should deepen collaboration with like-minded geographies like the European Union, Latin America, Africa and elsewhere to harness the power of OSAI. Given the limited financial resources of the middle powers, OSAI offers the only realistic pathway for replicating the success of Linux as a server operating system over better-funded proprietary operating systems like Windows NT and Unix. 

Conclusion

It may seem that India is far behind in the AI race. We significantly lag behind the US, which benefits from decades of research, almost half a trillion dollars invested in AI, and a concentration of leading AI companies with market capitalisation running into trillions of dollars. We are also behind China, which identified AI as a strategic priority early on, invested USD 120 billion in the sector, and has emerged as a world leader in AI research. 

However, as the examples of wheat and nuclear power illustrate, India has faced challenging situations in the past and responded by seeding national projects to ensure food security and indigenous nuclear technology. In crafting a similar trajectory in AI, open source can be a force multiplier that propels India from Tier 2 to Tier 1 status. 

Recommendations

Near Term (1-3 years)

1. Build an Open Source Program Office (OSPO) within the IndiaAI Mission. Fund it adequately to support the development of high-impact OSAI projects for India and the world. Task the OSPO with building an OSAI policy for India by building on the existing open source policy. Actively monitor the health of Indian OSAI projects on a regular basis. 
2. Mandate that mission-critical AI systems should be hosted in India on open source AI models. Depending on proprietary AI systems can pose severe risks for Business Continuity Planning due to vendor lock-in and denial of technology risks. This should therefore be avoided as far as possible.
3. For non-mission-critical applications, foreign-hosted large language models (LLMs) may raise concerns relating to surveillance, profiling, and data governance. Therefore, we must wean Indian users away from closed, foreign-hosted solutions to domestic solutions. To build trust in domestic AI service providers, strong protections against privacy violations and surveillance should be enforced.

Long Term (1-10 years)

1. Invest in deep academic research in India: India must aim to become one of the world’s top publishers of high-quality research papers on AI. To mitigate geopolitical uncertainties and protect against denial-of-technology regimes, we must develop the capability to create our own AI technologies. 
2. Collaborate with the EU, multilateral agencies, and the Global South to build a Global AI Commons: Given the hundreds of billions of dollars being invested by the U.S. and China in building AI infrastructure and capabilities, as well as the substantial headstart enjoyed by both countries, the only way Tier 2 countries can move to Tier 1 status is through deep collaborations with global open source communities, and by combining forces with like-minded partners. 
3. Elevate the role of OSS from a tactical to a strategic resource: Invest heavily in building vibrant open source communities and fostering the open source culture of collaboration. Open source should not be viewed through a zero-sum, geopolitical lens. Rather, it should be understood as a global commons that enables shared innovation and collective problem-solving.
4. Invest in shaping norms, standards and definitions: Given its historical position as a software services economy, India has been a rule-taker rather than a rule-maker. India’s membership in initiatives such as the open source RISC-V microprocessor consortium is a good example of an attempt to be at the rule-makers’ table. 
5. Leverage government purchases and the size of India’s domestic market effectively to grow India’s AI ecosystem: Public-sector procurement and India's large domestic market can play a critical role in nurturing indigenous AI capabilities, creating demand for domestic innovation, and supporting the growth of a sustainable AI ecosystem.

India's challenge is not simply to adopt AI, but to shape the institutions, technologies, and communities that will govern its future development. Just as strategic investments in agriculture and nuclear energy enabled India to overcome earlier periods of dependence, sustained investment in open-source AI can help the country build resilient capabilities while contributing to a more inclusive global AI ecosystem.