Google considers AI servers in India: minister

Ashwini Vaishnaw, India’s IT minister, said on May 11 that Google had agreed to seriously consider making AI servers in India after he raised the issue in a meeting with company executives last week.

Google designs chips such as TPUs but relies on partners including Quanta Computer and Foxconn to assemble server racks, while higher-level integration and TPU testing are largely done in the United States.

The remarks come as Google expands in India with a reported US$15 billion AI hub in the Indian city of Visakhapatnam, situated along the country’s eastern coast.

The plan could include up to 5 GW of data center capacity and as Vaishnaw said HP has begun making AI servers locally.

🔗 Source: The Economic Times

🧠 Food for thought

Implications, context, and why it matters.

India reworks PLI 2.0 for AI servers

India is revising the Production Linked Incentive (PLI) 2.0 scheme for IT hardware. The current model rewards incremental turnover and net sales, which fits poorly with AI servers because graphics processing units (GPUs) make up nearly 90% of costs and commercial high-performance GPUs are not made in India 1.
The government wants incentives to track value added in AI servers and advanced parts, rather than assembly volume 2.
That helps explain why India’s IT minister urged Google to make servers, GPUs, and chips in India during the foundation ceremony for its proposed US$15 billion AI hub in Visakhapatnam 3.

Google could help India build higher-value hardware

A local manufacturing commitment from Google could give domestic electronics makers a steady buyer in India’s AI server market 2.
That would back India’s effort to deepen its hardware base as more of the money in AI servers sits in premium chips and GPUs 4.
If that works, other global tech firms building infrastructure in India may copy the move, which could help the country make some of the hardware behind its cloud services.