How does India's AI strategy compare to those of the US and China

Introduction

India's artificial intelligence (AI) strategy represents a distinct third path in the global AI race, diverging from the U.S.’s private-sector-driven innovation model and China’s state-directed industrial dominance.

While the U.S. and China collectively control over 80% of global AI patents and 70% of advanced computing capacity, India's approach prioritizes scalable, socially impactful applications that are tailored to its demographic and economic realities.

This strategy has led to early successes in healthcare, agriculture, and governance, but it also exposes critical gaps in foundational research, semiconductor autonomy, and talent retention.

Strategic Frameworks: Divergent Philosophies and Objectives

United States: Private Sector Leadership with Defensive Controls

The U.S. maintains its AI supremacy through a decentralized ecosystem that combines academic excellence—producing 57% of elite AI researchers—with venture capital dominance ($67.2 billion in annual private AI investment) and military-civil fusion.

Recent policies under the "AI Diffusion Framework" prioritize defensive measures:

Export Controls

Restrictions on advanced semiconductors (e.g., Nvidia H100 chips) to China and over 140 countries.

Talent Retention

75% of U.S.-trained AI PhDs remain in the country, compared to India's 50% brain drain rate.

Compute Dominance

The U.S. has 5,200 petaflops of AI compute capacity—35 times India's 148 petaflops.

However, overreach in export controls risks alienating allies; 18 "trusted" countries receive unrestricted access, while India is classified as a tier-two country alongside Tuvalu and Vietnam.

This has accelerated Chinese inroads into emerging markets via open-source alternatives like DeepSeek-R1, which matches GPT-4’s performance at only 20% of the computational cost.

China: State-Led Industrial Policy with an Efficiency Focus

China's $137 billion annual investment in AI aims for global leadership through:

Centralized Planning

The 2017 Next Generation AI Plan coordinates five national champions (Baidu, Alibaba, etc.).

Data Advantage

China generates 25% of global data compared to India’s 9%, enabling breakthroughs in training efficiency.

Export Control Circumvention

Domestic production of 7nm semiconductors and algorithmic innovations reduce reliance on Western hardware.

This strategy has yielded tangible results: China files 2.5 times more AI patents than the U.S. and leads in industrial AI adoption. However, ethical concerns persist, with 92% of Chinese AI research lacking transparency protocols, versus 34% in the U.S.

India: Application-First Approach with Digital Public Infrastructure

India’s #AIForAll strategy, anchored in NITI Aayog’s 2018 framework, emphasizes:

Social Impact

Deploying AI in healthcare (72% district coverage), agriculture (18% yield increases), and multilingual governance.

Cost Innovation

Subsidized compute costs at $1.15 per hour, compared to the global average of over $2.50.

Sovereign Infrastructure

India has acquired 18,693 GPUs for domestic model training and invested $1.15 billion in the IndiaAI Mission.

This approach prioritizes developmental needs over geopolitical competition. India’s AI healthcare market, valued at $7.5 billion, along with initiatives like AgriStack digital twins, demonstrate localized problem-solving.

However, the country’s reliance on foreign LLMs for backend processing and its 0.4% share of global AI patents reveal structural dependencies.

Capacity Comparison: Talent, Infrastructure, and Innovation

India faces a talent deficit due to underfunded doctoral programs; only 0.08% of engineers pursue AI PhDs, compared to 4.2% in China.

Brain drain exacerbates these gaps: 7% of Indian AI graduates emigrate, primarily to the U.S., while China retains 94% of its talent through state incentives.

Compute and Data Asymmetries

Semiconductor Dependence : India imports 95% of its semiconductors, while China produces 70% domestically.

Data Governance

India’s National AI Data Marketplace has over 1 PB of datasets, but this is far behind China’s 450 PB of state-controlled repositories.

Open-Source Reliance

83% of Indian startups utilize Western or Chinese LLMs, compared to only 12% using domestic models.

The compute dominance of the U.S. and China enables them to train models with over 100 billion parameters, while India’s BharatGPT (currently under development) targets 30 billion parameters.

Sectoral Strengths and Strategic Vulnerabilities

Healthcare: India’s Asymmetric Advantage

Aarogya Maitri

AI for tuberculosis diagnosis has been deployed in 100,000 clinics, reducing detection time from 14 days to just 45 minutes.

Generative Drug Discovery

There is a 22% cost reduction in preclinical trials via domestically developed chemistry models.

These applications leverage India's scale but still depend heavily on foreign intellectual property; 87% of diagnostic algorithms use modified U.S. or Chinese architectures.

Agriculture: Precision vs. Productivity

ISRO’s Bhuvan

Satellite imaging with a resolution of 10 meters monitors 200 million acres.

Predictive Analytics

The system achieves 89% accuracy in monsoon forecasts to help with crop planning.

While these initiatives are impactful, they are narrow AI solutions compared to China’s comprehensive digital twins that integrate over 50 data streams across provinces.

Governance: Digital Public Infrastructure Edge

India’s 450,000 Common Service Centers provide AI-enabled services in 22 languages, contrasting with China’s social credit system. However, there are notable data localization gaps—only 34% of government datasets meet anonymization standards.

The Road Ahead: Strategic Recommendations

Immediate Priorities (2025-2027)

Talent Retention

Expand the PM Research Fellowship to retain 50% of AI PhDs domestically

Data Democratization

Launch National AI Data Marketplace with 1 PB+ India-specific datasets

Chip Diplomacy

Leverate QUAD partnerships to secure advanced semiconductor supply chains

Long-Term Vision (2030)

Sovereign AI Stack

Develop full-stack capabilities from silicon design (IN-Silicon) to multilingual LLMs (BharatGPT)

Global Standards Leadership

Champion ethical AI frameworks through Global South alliances

AI-Driver Productivity

Achieve 40% GDP contribution from AI-enabled services

Conclusion

Narrowing the Gap Through Strategic Asymmetry

India’s AI trajectory mirrors its pharmaceutical industry’s journey—initially dismissed, then dominant through cost innovation and scale.

While the U.S. and China battle for supremacy in foundational models and hardware, India’s $1.15/hour compute infrastructure and sector-specific solutions offer a viable third path.

Success requires sustained investment in talent (targeting 50,000 AI PhDs by 2030) and strategic partnerships to bypass semiconductor dependencies.

Current projections suggest India could capture 15% of the global AI services market by 2030, transforming from a peripheral player to an indispensable solutions hub—provided it navigates the precarious balance between technological sovereignty and global collaboration.