For eight years, a family car has been a reliable companion for school runs and grocery trips. Lately, however, visits to the authorised service centre have become more frequent as the vehicle begins to show its age. At around the same time, the family starts exploring replacement options through the same dealer – checking the trade-in value of their current car, comparing newer models on its website and enquiring about a test drive.
Individually, these interactions may not mean much. Put together with information the dealer already holds, such as the car’s age and service history, they can signal something more useful: The family may be ready for a new car.
A good salesperson might spot these clues and follow up. Artificial intelligence could do the same across thousands of customers simultaneously, helping the car dealer identify potential leads and respond to them in real time.
This is one way AI could help companies generate business, says Tri Pham, chief strategy and innovation officer at Tata Communications.
But for that seemingly simple AI-enabled sales process to work, different sources of data, computing power and communications tools have to connect with one another, and do so quickly, seamlessly and securely.
As companies move from AI experiments into incorporating AI into everyday operations, they need to look beyond the AI tools themselves and ask whether the digital infrastructure underneath them is ready, says Pham.
How networks can keep up
The current AI boom has “re-energised” the importance of networks and connectivity, says Pham.
In a 2025 survey by market intelligence firm International Data Corporation, 94 per cent of companies said their networks limit their ability to run large data and AI projects in some way. The survey covered 650 global enterprise technology leaders across Asia-Pacific, Europe and North America.
Part of the problem is that AI is developing rapidly but traditional network infrastructure – comprising hardware, software and connectivity services — has been comparatively static.
“Historically, network infrastructure is not as flexible or malleable as what we see in the cloud,” says Pham.
Tata Communications is making network infrastructure “cloud-like” – the same way cloud storage lets people pay for – and use – only the space they need, rather than buying a fixed hard drive upfront. This means companies can easily use what they need from their network and adapt as requirements change.
For example, the car dealer’s AI use might grow from simply answering customer questions to running AI agents that handle more complex tasks. Rather than buying new hardware and software at each stage, it can draw on what it needs from Tata Communications’ suite of digital solutions, which brings together network, cloud, interaction and Internet of Things solutions.
“Our customers are seeing that this will actually help improve their sales efficiency and be able to deliver potentially more revenues,” says Pham.
As companies start using AI, their networks must link the right data to the applications and computing resources that need it, securely and in line with regulations, says Tata Communications’ chief strategy and innovation officer Tri Pham.
PHOTO: TATA COMMUNICATIONS
Scale the network as AI use grows
However, upgrading the network is not necessarily the first step for a company preparing to use more AI.
“The challenge for most companies, even before thinking about the network or infrastructure, is thinking about where their data is,” says Pham.
Companies first need to understand whether their data is accurate and relevant, where it came from, where it can be stored and whether it can be moved.
The car dealer, for example, may have information spread across customer management systems, offices, connected devices, private data centres and cloud platforms.
If it was a multinational company, it may also have data in different countries, with different requirements governing where information can be stored and processed.
When the company starts using AI, its underlying network has to connect the appropriate data to the applications and computing resources that need it, while meeting security and regulatory requirements. This is what Pham describes as “network agility”.
Rather than requiring a company to buy an entire suite of components and services that may include parts they do not need, Tata Communications identifies the company’s business problem and provides the combination of solutions it needs.
Companies can also control how much they pay for each separate service. The car dealer may set a target of 1,000 text messages and 500 AI voice calls to customers, but pay for only what they use.
As a company’s AI needs evolve, capabilities can be changed, dropped or added.
For example, a company whose AI applications occasionally cause spikes in data traffic may not need to double its bandwidth permanently.
“What you don’t want to do is go from a 500MB plan to a 1GB plan only to discover that the extra capacity is unnecessary most of the time,” says Pham.
“What companies need is help to fine-tune their individual needs, in this case with their bandwidth,” he adds.
Tata Communications’ on-demand bandwidth allows companies to match network capacity more closely to actual data requirements, scaling it as their AI use changes and expands.
That flexibility matters because while many companies are still at an early stage of AI adoption, Pham expects this to change quickly over the next few years.
For companies preparing for that future, being AI-ready is therefore not just about having the right AI tools, but about maintaining underlying network agility to adapt as quickly as the technology does, says Pham.
Visit the Tata Communications website for more information.
