Data, applications, computing power: Get the network wrong, and AI adoption stalls

As companies transition from AI experiments to daily use, the supporting network infrastructure becomes crucial for success, warns Tata Communications’ chief strategy and innovation officer Tri Pham. Eight years ago, a family car was reliable for daily use, but as it aged, more maintenance visits occurred, similar to how AI systems evolve over time.

When the family considers replacing the car, various interactions with the dealer, such as trade-in value assessments and website research, reveal untapped potential – similar to how AI can identify business opportunities across customers. For AI-enabled sales processes to function effectively, data, computing power, and communication tools must seamlessly connect, Pham explains.

As companies adopt AI, they must evaluate whether their digital infrastructure can keep pace, as networks are often the bottleneck according to a 2025 International Data Corporation survey. Ninety-four percent of global enterprise technology leaders reported their networks limit their ability to run large AI projects. Traditional networks are less flexible compared to cloud solutions, which allows users to pay for only the resources they need.

Tata Communications is adapting its network infrastructure to be more "cloud-like," enabling companies to easily scale resources as their AI requirements change. This approach helps companies like the car dealer, who might initially use AI for basic customer interactions but later expand to complex AI agents. The first step for companies is often understanding their data's accuracy, relevance, storage locations, and portability.

For a multinational car dealer with data scattered across various systems and devices, network agility becomes essential. Pham describes this as connecting data to applications and computing resources securely and efficiently, while meeting security and regulatory requirements. Tata Communications tailors solutions to specific business needs, allowing companies to pay for only the services they use as their AI applications evolve.

As AI adoption accelerates over the next few years, companies must ensure their underlying network infrastructure remains agile to adapt quickly to technological changes.