Mint Explainer | Why Big Tech is rushing to build AI data centres across India – Crypto News

Why does AI need dedicated data centres?

AI is by far the biggest data user at the moment, primarily because it consumes and processes vast amounts of information to produce results.

Companies require large, at-scale data centre infrastructure to train large AI models, the largest of which are based on multiple trillions of data parameters. Even more challenging is making such AI models available without limitations and latencies to users worldwide.

The availability of these facilities closer to users makes it easier for applications to load and run. Beyond latencies, localizing a data centre also means that it is less prone to global geopolitical sanctions and disruptions—and with increasing reliance on AI for both businesses and public services, AI data centres are likely to emerge as critical infrastructure in India.

Whose needs would most such facilities serve?

Some of the biggest announcements, such as Microsoft’s $3 billion and Amazon Web Services (AWS)’s $6.8 billion, are designed to help the US-based Big Tech companies run their own applications, services and cloud operations. Google’s new $15-billion AI hub will also do the same.

However, local market infrastructure in AI typically prioritizes its usage in cloud-based AI applications that cater to the local market. For instance, Google’s AI hub in India, once operational, will host and run its Gemini AI platform.

OpenAI, which reportedly plans to set up a 1 gigawatt (GW) data centre in India, will also likely run its ChatGPT platform’s localized services, such as Indic language operations, out of the India facility.

Additional available data centre capacity can then be used to cater to the world, said Anushree Verma, senior director analyst for emerging technologies at research and advisory firm Gartner.

Which Indian states are leading the push?

Public data centre trackers indicate that Maharashtra is the most populous state in the country in terms of data centres, with 61 registered facilities. Tamil Nadu, Telangana and Karnataka are all close competitors, with 29, 25, and 24 data centres in each of the three states, respectively.

Andhra Pradesh, driven by Google’s announcement on Tuesday, is the latest entrant, but it is relatively behind the rest of the pack. Each of these states is offering favourable policies such as land and power subsidies as well as job creation-linked tax breaks and other incentives, to draw more data centres to their borders.

Are these data centres creating jobs?

On paper, each of the three Big Tech firms mentioned above has committed to creating jobs on a large scale. During his visit to India in January, Microsoft’s chairman and chief executive Satya Nadella said the company will upskill 10 million people in India with “essential skills”.

Google’s announcement implied that it would create 180,000 “high-value” jobs in Andhra Pradesh. Amazon, in January, said its investments in cloud and AI data centres would create over 81,000 jobs across Maharashtra and Telangana, where the facilities are being set up.

These jobs are emerging at a time when India’s $300 billion information technology (IT) services industry is experiencing a slowdown, resulting in a corresponding decline in hiring. The sector typically hires hundreds of thousands of people each year and is also critical to upskilling and training early-career engineers.

The data centres in question may somewhat offset the slowdown in IT hiring, industry stakeholders said. Moreover, jobs linked to data centres, cloud and AI services in India are likely to increase the median value of early-career jobs that engineering graduates pursue, thereby potentially benefiting India’s young workforce.

What’s the impact on local data centre providers and companies?

S. Krishnan, secretary at the IT ministry, said setting up hyperscalers and global data centres is a largely positive move for India’s industries. For one, startups working in the country on AI applications and services get access to the full stack of a Big Tech firm’s AI offerings as well as the latest data centre infrastructure in line with global standards.

For local data centre providers, increased competition also leads to a rising valuation of the overall data centre industry, which in turn can be beneficial in raising more capital to expand their own infrastructure and compete with Big Tech.

However, while Big Tech’s infrastructure largely caters to itself, existing local data centre providers, such as Yotta, Equinix, Princeton, NTT, and others, will cater to third-party clients and consumers, including Big Tech itself.

But what about power consumption and emissions?

States are offering policies that are helping data centre operators localize energy operations and emission management. For instance, earlier in October, the Maharashtra government announced that it would offer data centres with captive power generation and distribution licences in order to be able to cater to their own power needs.

The licences are largely centred on data centres generating their own green energy through renewable sources.

Uttar Pradesh, too, in its data centre policy published four years ago, prioritized self-generation of renewable power to run data centre operations in the state. Most service providers either purchase carbon credits to offset emissions generated due to the running of servers and cooling requirements or use renewable energy.