India needs a strategy reboot after US AI controls

India needs a strategy reboot after US AI controls

Former US president Biden’s policies on AI export controls and infrastructure have created challenges for India’s AI ambitions. The Framework for AI Diffusion requires India to fulfil regulatory compliance and authorisation requirements for AI collaboration with the United States. This, combined with the US Executive Order to build energy-intensive AI data centres that largely exclude India, complicates India’s efforts to develop its own AI infrastructure. India must negotiate bilateral agreements with the United States to bypass these restrictions and enhance energy capacity for AI operations. The US–India TRUST initiative is a good step towards an adjusted long-term strategy.

Just before leaving office, former US president Joe Biden passed two policies on artificial intelligence (AI). The policies are said to be the most vital releases since the October 2023 Executive Order on Safe, Secure, and Trustworthy Development and Use of AI as it marks a major shift in US AI export control policy and introduces a roadmap to build AI infrastructure domestically. The Trump Administration did not immediately revoke the two policies — unlike other AI decisions of the Biden Administration — reflecting an unexpected policy continuity. These policies will have a spillover effect onto other countries, including India, which will need to effect subtle short- and long-term policy shifts to cope with the new Cold War.

The Framework for AI Diffusion — the first of the two policies instituted by former president Biden – aims to implement export controls on AI diffusion. It seeks to protect US tech hegemony against efforts by China, Iran, Russia and North Korea to use high-tech US AI accelerator chips like Nvidia H100-equivalent chips and innovations in AI and computing infrastructure.

The Framework divides countries into three tiers based on their relative alignment with US interests and ability to prevent US adversaries from using their territory as indirect routes to buy US chips. Tier 1 countries include 18 US allies that do not face AI import restrictions. Tier 2 countries, including India, must create data centre authorisation programs or individual licenses to engage with US companies and implement safeguards against tier 3 country interference. Tier 3 countries such as Russia, China, Iran and North Korea face stringent import restrictions on US technologies.

The second of the two policies — the Presidential Executive Order on Advancing United States Leadership in Artificial Intelligence Infrastructure — aims to enhance the country’s domestic energy capacity so that it can build energy-intensive frontier AI data centres and computing infrastructure with the help of clean energy. The Framework mandates US companies keep 50 per cent of their compute capacity within domestic territory. This makes the Executive Order important in helping develop energy capacity to support this domestic focus and attracting investments in computing infrastructure.

Industry players have criticised these restrictions as they complicate efforts to build overseas AI infrastructure, while mandatory indigenisation increases costs.

It is not immediately clear why the United States put India on the tier 2 list given growing India–US technological collaboration. Some experts argue that the United States is concerned that Indian companies may be used to divert US resources to adversaries like Russia. This is the case for other countries on the tier 2 list — Israel and Singapore were both gateways for illegal imports of US technology to China.

These policy decisions create impediments for the India–US technology partnership, which had otherwise been improving since the signing of the Initiative on Critical and Emerging Technologies. Similarly to previous US restrictions on sharing critical technologies with India, such as nuclear reactors for civilian purposes and cryogenic engines for space rockets, recent decisions will raise questions about US credibility in helping build India’s technological capabilities.

The Framework challenges India’s efforts to build AI infrastructure and buy high-end chips. India should co-opt the US Framework by negotiating a bilateral arrangement that allows companies from both countries to cooperate without regulatory hindrances. The Framework itself facilitates this through a special review mechanism called the General Validated End User which gives Indian companies leeway to purchase restricted items for civilian and military — excluding nuclear — purposes.

During Indian Prime Minister Narendra Modi and President Trump’s meeting on 13–14 February 2025, both nations agreed to launch the US–India TRUST (‘Transforming the Relationship Utilizing Strategic Technology’) initiative, representing a significant step towards accelerating US AI infrastructure in India.

Continuous US–India engagement on critical technology has built enough confidence to effectively negotiate authorisation and licensing agreements as required for tier 2 countries. The latest joint statement by Modi and Trump reaffirmed cooperation on building next-generation data centres, advancing AI-focused computing and processor development and easing regulatory barriers.

But the Framework obstructs India’s ambitions to build local AI data centres by mandating companies based in tier 1 countries retain 75 per cent of their total AI computing power within tier 1 nations, while limiting their computing capacity in tier 2 countries to just 7 per cent. If India implements the data localisation provisions in its Draft Digital Personal Data Protection Rules 2025, it may be practically impossible for India to get tier 1 nations to establish data centres in India or even store Indian data within its physical boundaries.

India must pursue a bilateral arrangement with the United States for cross-border data flows and protection of its citizens’ data stored in US data centres, like the US–EU Privacy Shield. New Delhi must adopt a clear policy on data localisation amid the confusion created by the Data Protection law’s multiple iterations since 2018. India must also substantially enhance its energy capacity to accommodate AI data centre operators’ energy needs.

Bipartisan support within the United States for a nationalist AI policy to tackle competition with China means that India must adapt its strategy. The Trump Administration will likely continue — if not expand — the policy. India must pivot to navigate geopolitical challenges while balancing its technological progress and security needs.

Source: EastAsisForum