- APAC telecom operators are moving faster than peers in Europe and North America toward AI-native transformation
- Telcos face two major barriers in scaling AI beyond pilots: deep skills shortages and fragmented data architectures.
- AI could eventually disrupt router- and switch-centric network design
Telecom operators across Asia-Pacific are moving significantly faster on AI-led transformation than those in other regions, Google Cloud executive Ayush Sharma told Fierce.
“If you look at the APAC region, the telcos have leapfrogged even North America and Europe,” Sharma said. Sharma is currently Google Cloud’s GTM Head for Telecom, Media and Gaming for the APAC and Japan region, and was previously Chief Technology Officer (CTO) at StarHub, Singapore’s second-largest service provider.
Several telcos in the APAC region – including Singtel, SK Telecom and Indosat Ooredoo Hutchison (IOH) – are trying to transform themselves using AI. But they’re not all going down the same path. Sharma noted operators are pursuing a wide range of AI strategies.
While Japan is focusing on developing gaming and media use cases, Indonesia is focusing on AI infrastructure services like GPU-as-a-Service. “In India, we see more like cost leadership and effectiveness from a business model standpoint, not so much technology innovation,” Sharma added. “On the other hand, in China, we see technology innovation and AI being integrated, infused and new models being developed.”
Taking the lead
There are several factors that explain why APAC telcos are ahead of their counterparts in other regions.
To begin with, a number of leading operators, including Reliance Jio in India, True in Thailand and Rakuten in Japan, are part of large, diversified conglomerates. This makes it easier for them to develop and scale new digital and AI use cases by leveraging group-level assets across sectors such as retail, media, finance and enterprise, creating natural cross-vertical synergies.
In addition, many APAC telecom markets are highly price sensitive and characterized by relatively low Average Revenue Per User (ARPU). This puts sustained pressure on margins and pushes operators to adopt new technologies faster to improve operational efficiency and reduce operating costs. As a result, service providers in the region tend to be more proactive in deploying automation and AI-led tools across networks and operations.
Talent and data siloes remain core bottlenecks
However, as telcos shift from AI trials to scaled deployments, they face a major challenge in attracting talent. “The gaps are from a skills point of view, and that’s massive for telco[s]. Today, to attract somebody who’s worked for a hyperscaler or AI company to enter telcos is almost impossible,” said Sharma. Indeed, several players, including Axiata, have highlighted the skills problem faced by the telcos.
In addition, Sharma warned that without leadership commitment and operating model change, AI programs risk being mislabelled as technology failures.
“If you don’t have that the mindset and the drive from the board, these programs will run out of steam in a couple of years, and then that gets labelled as the technology adoption gone failure,” he said. “But the fact is it’s about your readiness and operating model and your culture of organization, not just about technology.”
On the technical side, fragmented and incomplete data foundations are holding back service providers from realizing their vision of becoming AI-native telcos. “AI cannot operate without a proper data strategy, without a proper data stack, tech stack,” he warned.
“In terms of customer data, billing data, etc., they have some [in a] data lake but it is not pervasive yet, and without a universal stack on the data, real AI native will not be possible,” added Sharma.
AI could disrupt traditional network design
Beyond operations and IT, Sharma believes AI could eventually reshape core network architecture itself. He suggested future networks could shift toward compute-heavy, AI-driven control planes.
“Today’s telecom networks run on fixed routing algorithms embedded in routers, switches and radio nodes. AI changes that equation because it can replace static algorithmic decision-making with adaptive, compute-driven intelligence,” Sharma said. “Over time, you could see networks rely less on purpose-built routing hardware and more on GPU-based, software-defined platforms running across cloud and edge. If that shift accelerates, it will be highly disruptive.”
He added that cloud- and AI-integrated satellite and low-earth-orbit connectivity models are early indicators of how disruptive that shift could become over the next several years.
Hyperscalers in APAC
As APAC telcos press ahead on AI, their relationship with the hyperscalers is evolving as well.
“Five years ago, telcos typically used hyperscalers selectively, mainly for specific virtualized workloads that couldn’t run on-prem due to regulatory, cost or risk considerations. Today, many of those constraints have eased,” Sharma said. “Cloud availability has expanded, pricing has become far more competitive, and most telecom workloads are now cloud-compatible.”
He continued: “For operators that want to become truly AI-native, cloud is no longer an optional delivery model, it has to be embedded into the core network and IT architecture rather than treated as an add-on productivity tool.”
APAC is a key focus area for most hyperscalers, including Google Cloud and Amazon Web Services (AWS).
Google Cloud has recently opened a new cloud region in Thailand as part of its $1 billion investment in digital infrastructure in the country. In addition, it announced an investment of $15 billion over the next four years to help build an AI region in Visakhapatnam in India. As part of this initiative, it is collaborating with Bharti Airtel, India’s second-largest service provider and Adani Connex to build a gigawatt-scale AI and data center hub in the city.
Asked how Google Cloud fights growing competition in the region, Sharma said the company’s “‘one Google’ approach — spanning AI models, platforms and applications — is what differentiates us, because most competitors either don’t have that full-stack product breadth or don’t have their AI capabilities tightly integrated into their broader platforms.”