The United States has made a bold move in the global artificial intelligence (AI) race with its new federal AI infrastructure policy. What appears to be a domestic initiative to establish AI data centres across federal lands, is actually a well-thought-out strategy to maintain America’s technological supremacy in this field. Other countries have also developed their own AI infrastructure strategies to face this competition. India faces resource constraints. The challenge for India, therefore, lies in adopting a strategy which will work without the deep financial reserves of the U.S. or China.
AI can be considered to be a technology in the mould of general purpose technologies (GPTs) that Jeffrey Ding has described in his GPT Diffusion Theory. It has a pervasive impact on different sectors, and requires a comprehensive infrastructure, including skill infrastructure, to thrive and diffuse.
Techno-nationalism
The U.S. policy aims at national security and AI leadership at its core. This is not just about building data centres. It is about building enough infrastructure domestically to ensure that research and development in this field can progress without any impediments. It can help strengthen America’s position as the global AI gatekeeper, if it can control the compute or resources layer of the AI technology stack. However, does not this raise important questions about the future of technological cooperation in an interconnected world? This throws to the wind the traditional concepts of market forces and comparative advantages which should be driving the allocation of resources globally. Are these to be superseded by national security concerns?
The answer is not straightforward. Techno-nationalism is here to stay, where a country’s worth is decided by its technological prowess. The U.S. has already imposed restrictions on the export of high-end AI chips to China. This highlights the geopolitical dimensions of AI infrastructure development.
The world will see a lot of this competitive one-upmanship in the garb of narratives of self-sufficiency and technological leadership. It would be better for nations to do a comprehensive cost-benefit analysis of such decisions to be able to harmonise national interests with global partnerships.
China also has invested heavily in domestic AI chip manufacturing and government-backed AI research. Companies such as Huawei and the Semiconductor Manufacturing International Corporation (SMIC) are at the forefront of developing indigenous alternatives to Nvidia’s AI chips, ensuring that China is not entirely dependent on western technology. A lot of this is driven by creative insecurity brought on by the tech and trade wars unleashed by the U.S. In addition, China, too, has aggressively been expanding its data centre networks, integrating AI computing with its broader push for digital infrastructure dominance under the Belt and Road Initiative. These are different models of doing the same thing — establishing AI supremacy. Unlike the U.S., which is leveraging public-private partnerships, China’s model is deeply state-driven, with massive subsidies and policy support ensuring rapid progress. One could argue that both are market distorting interventions.
The European Union is also taking AI infrastructure seriously but with a focus on ethics, regulation, and sustainability. They have been investing in sovereign cloud infrastructure to reduce reliance on foreign multi-national companies. Additionally, it is actively promoting open-source AI models. The aim is to keep AI development transparent and accessible to smaller businesses and researchers.
The issue of sustainability
The U.S. has also dabbled with such an approach. Consider this policy’s environmental stance. There is a commitment to power next-generation data centres with clean energy. This is not how traditional infrastructure development happens. However, the Trump administration’s track record on climate issues has not been great. It has withdrawn from the Paris Agreement, among other such reckless actions. Hence, this policy’s environmental stance may indeed change in the future, and needs to be taken with a huge pinch of salt. Recent studies indicate that energy consumption for running AI infrastructure will most likely rise exponentially. Innovations such as the DeepSeek model, which showcased a lot of efficiency improvements, do give a lot of hope that this may not be the case entirely. However, it would be prudent to err on the side of caution, and consider sustainability options.
The U.S. policy also explicitly addresses community interests and worker benefits. It acknowledges that technological advancement must not disregard broader societal interests. If this approach could create new tech hubs beyond Silicon Valley, it could potentially help in spreading the benefits of the AI revolution more equitably across the country. However, the Trump administration did also recently rescind a previous executive order which mandated a bunch of disclosures on AI developers to share the results of safety tests with the U.S. government. The balance between innovation and safety regulation is tough to achieve.
What India must do
For countries such as India, these diverse approaches offer valuable lessons. The model of utilising government lands for AI infrastructure development through public-private partnerships, by leasing out federal lands, could be particularly relevant, instead of, say, government procuring GPUs. India cannot afford to match the U.S. or China in sheer financial might. It needs to adopt a strategic, targeted and collaborative approach to AI infrastructure development, which leverages its strengths.
One of India’s biggest strengths is its vast talent pool in AI and software development. India does not yet have a robust AI hardware industry. However, it is already a major hub for cloud computing services, and software engineering. The presence of global AI firms in India, combined with domestic giants such as TCS and Infosys investing in AI, could provide a foundation that can be leveraged. However, for this potential to be realised, India needs to develop the necessary computing infrastructure to support AI innovation at scale.
A major challenge is India not having high-end AI chips and large-scale computing power. India could partner with U.S. and European firms to establish joint AI computing hubs in India. It helps India to participate in the global AI economy without needing to invest billions in semiconductor fabs.
The next important factor is India’s energy constraints. AI data centres require a huge amount of power. India, at its level of economic development, has to make hard choices between expanding energy access for industrial growth and dedicating power to AI computing. A viable path forward is to integrate AI data centres with renewable energy projects, balancing both imperatives, and creating a sustainable AI growth model.
Given India’s resource constraints, India should prioritise a few key AI hubs that combine computing power, talent, and industry partnerships, rather than attempting a nationwide rollout. The government’s IndiaAI Mission has got this part right, as it already recognises the need for national AI supercomputing resources. But this initiative must be expanded with stronger incentives for private sector participation. Creating AI-specialised free trade zones with relaxed regulations on AI chip imports and software exports could also help position India as a competitive AI hub.
AI infrastructure is the key factor in the global race for AI supremacy. Play to one’s strengths. Acknowledge the constraints. Collaborate to be in the race.
Arindam Goswami is a Research Analyst in the High-Tech Geopolitics Programme at The Takshashila Institution, Bengaluru
Published – March 01, 2025 02:00 pm IST