As telecom operators scramble to define their place in the artificial intelligence (AI) value chain, one question continues to hang over the industry: what happens when the hyperscalers arrive?
For many operators, the prospect of competing against the likes of OpenAI, Google, Microsoft and Anthropic appears daunting. Yet for Veon, whose markets span countries including Kazakhstan, Pakistan, Bangladesh and Ukraine, the rise of global AI giants is not viewed as an existential threat.
Instead, the operator sees itself occupying a different layer of the AI ecosystem altogether.
“We’re at the point where AI is becoming the next accelerator of engagement,” said Lasha Tabidze (pictured), Chief Digital Operations Officer at Veon, during a recent media roundtable in London.
The company has spent the past several years transforming itself from a traditional telecoms operator into what Tabidze describes as “a digital services and enterprise company, which also happens to have the telecom licence”.
That transformation now sits at the heart of Veon’s AI 1440 strategy, an evolution of its earlier Digital Operator 1440 programme. The name refers to the 1,440 minutes in a day, with the company aiming to embed AI into customers’ daily lives across finance, healthcare, education and entertainment.
The strategy is already being deployed at scale. Veon operates across markets with a combined population of around half a billion people. Across those markets, roughly one in three people use one of its telecoms services, while one in two use one of its digital platforms.
Distribution matters more than models
While much of the AI industry’s attention remains focused on increasingly powerful large language models (LLMs), Tabidze argues that the real battle has shifted elsewhere.
“Nobody wakes up in the morning thinking about which GPT model they’re using,” he said. “People think about healthcare, education, entertainment and financial stability.”
That belief underpins Veon’s conviction that AI deployment will ultimately prove more valuable than AI development alone.
Rather than trying to build a direct rival to ChatGPT or Gemini, the company has focused on creating local language models and AI-powered services that sit on top of existing global foundation models.
In Kazakhstan, Veon developed Kaz-LLM, a locally trained language model designed around Kazakh language, culture and context. Similar initiatives are underway in Bangladesh, Pakistan and Ukraine.
The approach reflects a pragmatic view of the AI landscape.
“I don’t think there’s any value in competing against hyperscalers,” Tabidze said. “They have resources, they are doing this for the world, they’re doing huge investments.”
Trying to match that investment would be impossible, he argued. “You cannot start discussing putting in 1,000 GPUs when a hyperscaler is discussing 200,000 GPUs in a new data centre.”
Instead, Veon uses global foundation models as a base layer before adding local data, language capabilities and domain-specific functionality.
For the operator, the competitive advantage lies not in building the world’s biggest AI model but in ensuring that AI works effectively for a farmer in rural Bangladesh, a student in Kazakhstan or a small business owner in Pakistan.
Local AI for local markets
A recurring theme throughout the discussion was the importance of context.
Tabidze argued that translation alone is insufficient for many AI applications. Local language models can better capture cultural references, historical context, regulatory requirements and linguistic nuances that global models may overlook.
That is particularly relevant in Veon’s markets, many of which have languages that receive far less attention from major AI developers.
The company views this as both a commercial opportunity and a way of preventing AI from widening the digital divide.
“The next billion users of AI will not be coming from the West,” said Tabidze. “They will be coming from countries like Pakistan, Bangladesh and India.”
For many users in these markets, smartphones represent their primary – and often only – gateway to the internet. Voice interfaces could also become increasingly important where literacy rates remain lower than in developed markets.
The objective is to make AI accessible and affordable enough for mass-market adoption.
That affordability challenge is significant. In some of Veon’s markets, average monthly telecom spending remains below US$2, making Western AI subscription models difficult to replicate.
Telecom’s hidden AI advantage
Veon believes operators possess a major advantage that many in the industry have historically underestimated: distribution.
“Telecoms are the cheapest distribution network for any digital product,” Tabidze argued.
The logic is straightforward. Operators already maintain trusted relationships with millions of customers, understand usage patterns and possess established billing systems.
Those capabilities become particularly valuable when introducing AI services.
Today, more than 2.5 million Veon users actively use AI agents within its platforms. The company’s digital ecosystem now reaches more than 240 million active users over a 90-day period, including around 70 million who are not telecom subscribers at all.
That reach allows Veon to integrate AI directly into existing digital services rather than relying on standalone chatbot applications.
The company has already deployed AI across financial services, healthcare and education. In Kazakhstan, it is also testing AI commerce capabilities that allow users to search for products, make purchases and complete payments through conversational interfaces.
Meanwhile, Veon’s fintech operations now serve around 60 million mobile financial services users.
Growth increasingly driven by digital services
The operator’s confidence in its AI strategy is reflected in its financial ambitions.
According to Tabidze, digital businesses accounted for less than 7% of Veon’s revenues only a few years ago. Today they contribute roughly 25%.
In the first quarter, digital revenues grew by nearly 58% year-on-year in US dollar terms, significantly outpacing the company’s telecoms business, which also delivered double-digit growth.
Veon openly expects digital services to account for half of total revenues by 2030.
The company sees AI as the primary catalyst for reaching that milestone.
As AI becomes embedded into customer-facing applications, Veon expects higher engagement, greater personalisation and stronger monetisation opportunities across its digital portfolio.
Sovereignty, trust and the future
The rise of sovereign AI also plays into Veon’s strategy.
Across many markets, governments are becoming increasingly focused on where data is stored, how AI models are trained and who ultimately controls critical digital infrastructure.
Tabidze views this less as a protectionist trend and more as a matter of digital independence.
Trust, he argues, will become one of the most valuable assets in the AI era.
As AI agents take on more responsibility – from financial transactions to healthcare support and e-commerce purchases – users will need confidence in the systems handling their data and making recommendations.
For operators, that could create an opportunity to leverage decades of customer trust and infrastructure ownership.
Whether telecoms can successfully capitalise on that opportunity remains one of the industry’s defining questions.
Veon, however, appears convinced that the future of AI will not be decided solely by whoever builds the largest model.
It will also depend on who can deploy that intelligence most effectively, make it affordable and place it in the hands of millions of users.
On that front, the company believes telecom operators still have a role to play.
