International Relations15 September 20265 min read
India proposes BRICS reform roadmap as China offers open-source AI community
The Indian Prime Minister asked BRICS leaders at the 18th BRICS Summit in New Delhi in September 2026 to frame 10 global governance reform proposals, to be built into a BRICS Reform Roadmap by the 2027 summit. At the same summit, China proposed a BRICS open-source AI community, an offer that sits uneasily with India's push for autonomy in artificial intelligence.
The UN Secretary-General separately said that "distorted power relations in the world make the Security Council non-operational", and reaffirmed the need to address the democratic deficit in multilateral institutions.
What India's roadmap contains
The roadmap is organised around three themes, the 3Rs: Representation, Responsiveness and Rule-making in emerging technologies. India called for turning a "pyramid of privilege" into a "platform of partnership", saying the Global South faces global crises at the forefront yet remains excluded from key decisions.
On representation, India said UN Security Council (UNSC) reform cannot be delayed and that text-based negotiations on expansion must carry a definite timeline and clear outcomes. Voting power and leadership roles at the IMF and World Bank must, it said, reflect contemporary economic realities.
On responsiveness, it proposed a Seafarers' Emergency Support Network linking maritime authorities and missions for rapid distress alerts, medical aid and evacuation of maritime workers. On rule-making, it said equal participation is vital in framing rules for AI, cyberspace, outer space, biotechnology and critical minerals, and that young diplomats from the Global South should get legal, practical and negotiation expertise.
To keep momentum from stalling with the annual rotation of the Presidency, India also proposed a BRICS Continuity and Implementation Mechanism. It would use the BRICS Troika (the current, previous and incoming chairs) and a secure digital repository to track initiatives, deadlines and implementation status. India said the effort rests on dialogue, consensus and cooperation, and is not a counter-bloc against any country.
Why reform is hard
The 11-member BRICS accounts for about 40% of global GDP and nearly half the world's population. The obstacles to reform are well entrenched:
- The UNSC reflects the post-World War II order. The G4 (India, Brazil, Germany, Japan) seek permanent seats, and Africa's demand for two permanent seats is set out in the Sirte Declaration and the Ezulwini Consensus. The "Uniting for Consensus" group opposes new permanent seats, with Pakistan opposing India's bid.
- P5 vetoes used to shield allies have paralysed the Council, as with Russia on Ukraine and the US on Gaza.
- An unwritten understanding reserves the World Bank presidency for an American and IMF leadership for a European.
- The US has blocked WTO Appellate Body appointments since 2019, crippling dispute settlement.
- Unilateral measures such as the EU's Carbon Border Adjustment Mechanism (CBAM) and the US Inflation Reduction Act (IRA) burden developing economies.
The steps listed to address these include time-bound, text-based negotiations on expanding the UNSC in both permanent and non-permanent categories, quota realignment at the IMF and World Bank, and ending the WTO Appellate Body deadlock while safeguarding Special and Differential Treatment.
China's AI offer and India's dilemma
China offered to pioneer a BRICS open-source AI community covering the development and application of large language models, AI seminars and training courses, and an open ecosystem backed by a unified digital cloud platform.
The proposal is a software-led route around Western export controls on advanced AI chips. It mirrors Google's Android strategy: open-weight models such as Alibaba's Qwen and DeepSeek would be distributed freely in emerging markets, where developers can download, modify and build on them, as an alternative to costly proprietary Western systems.
India advocates a "third way" in AI governance based on strategic autonomy, democratic values and privacy-preserving infrastructure. Joining could shift its dependence from US proprietary models to Chinese ones, whose data and training practices may lack transparency. India's digital architecture stresses democratic values and human rights, whereas China's AI regime requires models to align with state objectives. Free foreign models could also undercut Indian model builders such as Sarvam and BharatGen.
Gaps at home
India's own capacity is limited. Data centre capacity of 1.6 GW in mid-2026 is projected to reach 6 GW by 2029, needing about USD 110 billion. Data centres use roughly 150 billion litres of water and 0.5% of national electricity annually (CEEW, 2025), and Mumbai holds up to 50% of stock. India imports 90 to 95% of its semiconductors. The India Semiconductor Mission has drawn over USD 20 billion, but facilities focus on mature nodes (28nm to 110nm) and packaging, leaving India reliant on foreign foundries such as TSMC for sub-5nm chips.
The Digital Personal Data Protection Act, 2023 uses a negative-list approach to cross-border data flows, which gives the government wide discretion and delays foreign cloud investment. The Copyright Act, 1957 leaves open questions on fair use for LLM training and ownership of AI outputs.
The remedies listed are sovereign AI under the IndiaAI Mission (₹10,300 crore), domestic GPU clusters through public-private partnerships, and extending the Digital Public Infrastructure model to AI as a democratic alternative to China's offer. They also include deeper work on the US-India initiative on Critical and Emerging Technology (iCET) and the Quad Semiconductor Supply Chain Initiative, and use of the G20 and the Global Partnership on AI to push a multi-stakeholder governance framework.