Tata Communications has started discussions with enterprise customers to secure deals for its AI Cloud in partnership with chip giant Nvidia and plans to kick off trials soon.
“We are working hard to launch the AI Cloud and expect to be ready for trials towards the end of the year. We’re excited about how it’s shaping up. Yes, we have begun discussions, and there’s a lot of interest, not just in India but internationally as well,” AS Lakshminarayanan, managing director and chief executive of Tata Communications, told Moneycontrol in an interaction.
The company has started seeing interest from enterprises in the manufacturing and retail sectors for edge inference. “Banks are more cautious, especially in customer-facing processes, as they want to ensure trust in AI decisions. If something goes wrong, they can't say things like AI refused the mortgage for the customer. They have to be accountable for it. However, they are exploring AI in back-office functions, which are less risk-sensitive,” he said.
With the AI Cloud, Tata Communications will also consider fetching customers from overseas markets in addition to its core audience in the domestic market.
Tata Communications is building an AI Cloud powered by the next-generation Nvidia GH200 Grace Hopper Superchip. In September, Nvidia announced an extensive collaboration with Tata Group to deliver AI computing infrastructure and platforms for developing AI solutions.
The company is not only about buying the graphics processing unit (GPU) but needs to do the engineering work to get the architecture right and set up the data centres. It also builds the platform layer on the infrastructure layer.
The top executive said that most enterprises' current AI spending is focused on training rather than inferencing, but that will shift over the next few years. “While enterprises are still experimenting with AI, we expect gradual adoption as they move from proof of concept to full-scale deployment,” he added.
Enterprises in India are experiencing the AI Cloud and are willing to provide proof of concepts. However, he said that the PoCs themselves will not be at a big scale.
“We believe that it will be a gradual take-up, initially by people who want to train models. Inferencing will be small. In a few years, inferencing will become big. We are very excited about AI… The enterprise take-up is going to be more gradual,” he added.
He said enterprises would first want to see the business case, determine whether the model is trustworthy, and ensure that the data they give to the model is kept within their four walls or, if they are not, train a model that competitors can use.
“We are in the very early stages of assuring all of these for customers. While AI has the potential to do what the internet did two decades ago, it also has the same potential. As with the Internet, it took time; this will also take time,” he added.
Lakshminarayanan said that while the company has expected deals from enterprises for its AI Cloud since day one, it doesn’t think they will be big-ticket. “We will secure customers. We'll have to see the size of all of that…I don't know whether I can claim that we will secure a multi-million deal or something with a couple hundred crore rupees. But we will have the deals, and it will be a journey,” he said.
The company will also launch a multi-cloud network product before the end of the year, which will allow it to serve a full stack of customers who want to connect to multiple clouds seamlessly, like a SaaS product.
Discover the latest Business News, Sensex, and Nifty updates. Obtain Personal Finance insights, tax queries, and expert opinions on Moneycontrol or download the Moneycontrol App to stay updated!
