In order to clinch partnerships for its AI Cloud in collaboration with semiconductor giant Nvidia, Tata Communications has begun negotiating with business clients. The company also intends to launch trials shortly.

"We anticipate being prepared for testing by the end of the year, and we are working hard to deploy the AI Cloud. We're thrilled with how things are developing. In an interview with Moneycontrol, AS Lakshminarayanan, managing director and chief executive of Tata Communications, stated, "Yes, we have started talks, and there is a lot of interest, not just in India but internationally as well."

Businesses in the manufacturing and retail industries have begun to express interest in edge inference, according to the business. "Because they want to guarantee that AI choices are trusted, banks are being more careful, particularly in activities that involve customers. They cannot claim, for example, that AI denied the customer's mortgage if anything goes wrong. They must take responsibility for it. They are investigating AI for less risk-sensitive back-office tasks, though," he stated.

In addition to its main audience in the domestic market, Tata Communications will think about attracting clients from other markets using the AI Cloud.

The next-generation Nvidia GH200 Grace Hopper Superchip is being used by Tata Communications to run an AI cloud. Nvidia and the Tata Group established a broad partnership in September to provide AI processing platforms and infrastructure for creating AI solutions.

The business must conduct technical work to get the architecture correct and set up the data centers in addition to purchasing graphics processing units (GPUs). On top of the infrastructure layer, it also constructs the platform layer.

According to the senior executive, most businesses currently spend more on training than inferencing when it comes to AI, but this will change in the coming years. "Even though businesses are still testing AI, we anticipate a slow uptake as they proceed from proof of concept to full-scale implementation," he continued.

Indian businesses are experimenting with the AI Cloud and are prepared to offer proof of concept. The PoCs themselves, he added, will not be on a large scale.

We think that those who wish to train models will be the first to adopt it gradually. There will be minimal inferencing. Inferencing will gain popularity in a few years. We are ecstatic about AI. "Enterprise adoption will be more gradual," he continued.

According to him, businesses would first want to evaluate the business case, assess the model's reliability, and make sure the data they provide is kept inside their walls or, if not, used to train a model that rivals might use.

"Assuring all of these for clients is still in its very early phases. AI has the same potential as the internet, but it can also achieve things that the internet did twenty years ago. This will take time, just like the Internet did," he continued.

Although the company has anticipated business partnerships for its AI Cloud since day one, Lakshminarayanan stated that it does not anticipate them to be expensive. "We'll get clients. We'll have to wait and see how big that is. With a few hundred crore rupees, I'm not sure whether I can say that we'll land a multi-million dollar transaction or whatever. However, it will be a road, and we will have the deals," he stated.

Before the year is out, the business will also introduce a multi-cloud network product that will enable it to provide a comprehensive stack of services to clients who wish to connect to different clouds in a seamless manner, similar to a SaaS product.