Archive · 6 Sep 2026
India Should Avoid China’s AI Price War, Pursue Value
China is making some AI services very cheap.
This could help more people and companies use AI.
India also has cheaper computer power and is building more data centres.
But cheap computers are not enough because AI needs electricity, cooling, chips, data, and skilled workers.
India could become a place that runs other countries’ AI without owning the important businesses.
The article says India should build AI for healthcare, farming, education, and public services.
These tools should work for India’s many languages and different communities.
The goal should be useful AI that creates lots of economic value, not simply the lowest price.
China is using low AI prices to accelerate adoption and expand influence across the technology ecosystem.
India has onboarded more than 38,000 GPUs through 14 providers at subsidised rates averaging about Rs 65 an hour.
Indian data-centre capacity increased from roughly 375 MW in 2020 to around 1,500 MW in 2025.
The article warns that cheap infrastructure alone could leave India hosting foreign intelligence while others own platforms, intellectual property, and customers.
It urges India to focus on population-scale AI applications and measure how much economic value each GPU creates.
- Who
- China, India, Chinese AI companies such as DeepSeek and Alibaba, and American frontier-model companies are central to the discussion.
- What
- The article argues that India should not copy China’s low-price AI strategy, but should pursue useful, high-impact AI and stronger ownership of platforms and applications.
- When
- The article cites infrastructure figures for 2020 and 2025 and discusses current AI pricing and competition.
- Where
- The competition involves China, India, the United States, and global AI markets.
- Why
- The author argues that price alone does not capture AI’s full economics and could repeat India’s earlier outsourcing pattern, while value-focused applications could create greater national benefit.
Price-Led AI Strategy
Value-Led AI Strategy
India’s competitive advantage
Price-Led AI Strategy
India could use low-cost compute and infrastructure to attract AI workloads, encourage experimentation, and become a cost-efficient global hub.
Value-Led AI Strategy
India should avoid becoming merely a low-cost host for other countries’ intelligence and instead develop distinctive applications and platforms.
What should be measured
Price-Led AI Strategy
Lower model and compute prices can expand adoption, reduce vendor lock-in, and make previously uneconomical applications viable.
Value-Led AI Strategy
The more important measure is how much economic value and productivity each GPU creates, not simply how cheaply it operates.
Strategic model
Price-Led AI Strategy
China’s approach emphasizes reducing costs, increasing adoption, building scale, and expanding ecosystem influence.
Value-Led AI Strategy
India should build on its digital public infrastructure, linguistic diversity, and frugal innovation to create population-scale AI in areas such as healthcare, agriculture, education, and government services.
- DeepSeek initial price
- DeepSeek’s V4 Pro initially cost $0.87 per million output tokens.
- DeepSeek peak-hour price
- Its peak-hour pricing later rose to $3.96 per million output tokens.
- Indian GPUs
- More than 38,000 GPUs have been onboarded through 14 providers.
- Subsidised compute
- Average subsidised access in India is about Rs 65 per hour, roughly one-third of the global average.
- Data-centre capacity
- Capacity increased from about 375 MW in 2020 to around 1,500 MW in 2025.
- Data-centre power costs
- Industry estimates place electricity at roughly 40% to 60% of data-centre operating costs.
- IndiaAI Mission
- The mission has a Rs 10,370-crore outlay targeting compute, foundation models, datasets, startup financing, and AI talent.