Strategic Vulnerabilities in Foreign AI Dependence
India’s increasing reliance on foreign-owned artificial intelligence models and technology infrastructure poses potential strategic risks as access to advanced technologies becomes more influenced by geopolitical factors, export controls, and national interests. This assessment comes from a report titled "India Needs a Sovereign AI Revolution" released by consulting firm Primus Partners.
The report argues that the nation must transition from being a large-scale adopter of global AI technologies to developing sovereign capabilities across the entire AI stack. These capabilities include compute power, models, data, infrastructure, talent, and applications. While the report acknowledges that global technologies will continue to play a role, it emphasizes that critical infrastructure, public services, and strategic sectors must not become dependent on AI systems outside of India’s control.
Devroop Dhar, co-founder and managing director of Primus Partners, stated that AI sovereignty is not about operating in isolation. Instead, it involves ensuring that India has the capability to build, control, and adapt the technologies underpinning its critical systems. Dhar noted that investing in indigenous intellectual property, compute, talent, and research and development (R&D) today is necessary to prevent access to frontier AI from becoming a strategic vulnerability in the future.
Unlocking Private Research Through CSR Amendments
The report identifies a significant gap in India’s current R&D investment levels. According to Primus Partners, India’s gross expenditure on R&D currently stands at approximately 0.84% of GDP, which is significantly below the global average of 1.92%. To address this disparity, the paper calls for greater private-sector involvement in R&D, particularly in priority fields such as AI, quantum technologies, and semiconductors.
Primus Partners points to India’s mandatory Corporate Social Responsibility (CSR) framework as a strategic lever to bridge this gap. Currently, companies are required to spend 2% of their net profit on CSR activities. However, the report highlights that under Schedule VII of the Companies Act, CSR funds can be directed toward R&D only via public institutions such as the Council of Scientific and Industrial Research (CSIR), the Department of Science and Technology (DST), and the Indian Institutes of Technology (IITs). Furthermore, this specific spend receives no tax deduction under Section 35 of the Income Tax Act.
The report alleges that widening CSR eligibility to cover in-house R&D in priority areas and frontier technologies would unlock a significant private pool for research. It suggests that an amendment to Schedule VII and Section 35 can easily enable this change. By allowing companies to direct these funds toward their own internal research initiatives, the government could stimulate innovation without direct fiscal outlay.
According to Primus Partners, redirecting even 10% of India’s annual CSR pool could create a Rs 4,000 crore kitty every year for in-house R&D. Over a five-year period, this approach can provide in excess of Rs 20,000 crore for research in frontier technologies, which the report describes as a potential game changer for the sector.
Adopting Model-Agnostic Approaches
To mitigate risks associated with vendor lock-in, the report recommends that businesses and government systems adopt a model-agnostic approach. This strategy ensures that applications can run on different AI models rather than being tied to a single provider. Primus Partners argues that this flexibility will allow organizations to adapt as the AI ecosystem evolves and reduce the risks associated with dependence on one technology provider.
Retaining Global Talent
The report also calls for greater efforts to bring Indian researchers and AI professionals working overseas back into the country. It suggests providing targeted support for research, relocation, and leadership opportunities to encourage this return migration. Building the ability to develop, control, and scale critical AI capabilities domestically is described as essential for the country’s technology competitiveness, economic resilience, and strategic autonomy.