Yotta to unveil new Greater Noida, Guwahati data centres in 2024

“The Yotta-D1 data centre is 85% full right now, and we’re now building D2 and D3 already. The full capacity will have six full-scale buildings. Construction is currently ongoing for our D2 and D3 facilities, and D2’s racks will go live in July or August itself," Gupta said.

The executive added that the Uttar Pradesh government has offered critical sops to Yotta to facilitate its data centre operations in the National Capital Region—the first such facility for Yotta in north India. “We’ve had to set up a 220kV substation on-campus to fulfill our power consumption requirements. It has taken us 18 months to lay down the requisite power lines for this, but we’ve been supported by the UP government to get it."

Yotta is also in process of expanding its tier-II market presence, but is doing so in a staggered way as demand for such facilities have not yet scaled up. “People are still trying to find the real use cases for ‘edge’ data centres. So far, edge use cases have been concentrated on content-driven operations such as Facebook and YouTube, because they need to put their caching infrastructure closer to users. These companies are reaching out to local ISPs and installing data caching instruments."

“But, the real use cases that we’re still chasing is that, with the advent of 5G, a lot of IoT and AI-driven use cases will come up, such as agricultural drones needing real-time data caching for offering analytics in fringe markets. These use cases haven’t really come up maturely, even though some conversations have happened in this regard. Real-world deployments are yet to be seen, which is what will fuel our edge expansion plan," Gupta said.

Talks of the first such facility, Gupta added, have happened in Guwahati. “In Guwahati, we’ve just signed a contract with a major customer. Now, we’ll start working on the facility there, and it’ll come on in around nine or 10 months," he said. Yotta has also identified Bhubaneswar and Nagpur as other such markets, among seven regions where the company has purchased land for expanding data centre capacities.

Data centre investments in India in the first half of this year reached $21.4 billion, as per a 5 December report by real estate consultancy firm, CBRE. While digital transformation and increasing adoption of artificial intelligence (AI)-driven data operations were key factors fuelling this growth, industry veterans said a regulation-driven boost to localization has not happened at scale—with the Digital Personal Data Privacy (DPDP) Act not explicitly recommending localization of data.

Gupta said most edge data centres will offer fairly simplified structure for the operator to be able to scale. “You cannot treat edge data centres like hyperscale facilities. Our edge data centre blueprints are of 2MW capacity buildings, and are based on small parcels of 3-acre lands. This way, cost and time of building these facilities are optimized. Because we’d eventually want to build hundreds of such data centres, it isn’t always possible to innovate in design. You need a standardized design to scale them. Our idea is that after the first three such facilities, our expertise will help us build and operationalize one such data centre within as little as four months’ time," he said.

However, despite multiple companies planning such expansions, India continues to remain stretched for ample data centre capacity. A Colliers report from 5 November pegged India as the 14th nation in the world in terms of overall capacity, with 151 data centres. This, Gupta added, has stalled AI development in the country due to lack of ample infrastructure.

“AI-driven data demand was already there, but India so far did not have infrastructure to locally build and develop AI models. That’s where our opportunity is right now, because of which we at Yotta build our own native full-stack cloud platform as an alternate offering to AWS, Azure and GCP. This platform is going live in February, and demand is being driven by the ChatGPT-boosted craze," he said.

He added that while startups are making increasing efforts to build large language models (LLMs), a lack of available infrastructure and exorbitantly expensive cloud platform and data processing costs with hyperscalers such as Amazon Web Services, Microsoft Azure and Google Cloud Platform remain key challenges that the industry needs to address.