From connecting users to hosting computing power, Philippine telecom giant Globe bets on AI data centers! Targeting the next billion-dollar business.

According to Zhihui Financial APP, AI agents are evolving from simply answering questions to continuously executing tasks, shifting growth opportunities from chip and cloud computing service providers to large-scale AI data center operators that offer comprehensive AI computing infrastructure, power resources, cooling, and high-speed network connection systems. Globe Telecom Inc, the second largest telecom operator in the Philippines, is betting on this transformation, seeking to develop its data center joint venture into its next billion-dollar business: the company plans to invest at least $1 billion in capital expenditures between now and 2027, mainly for continued expansion of its data business; Meanwhile, GCash’s parent company Mynt—which is also Globe Telecom’s fintech subsidiary Mynt Inc.—is planning to raise about PHP 60 billion through an IPO, potentially setting a record for the Philippine stock market.

The growth strategy of telecom giant Globe concurrently covers digital finance and computing power infrastructure. Globe’s investment logic in AI infrastructure can be summed up as competing for “AI computing landing rights”—combining reliable power supply, deliverable data centers, high-density cooling, and cross-border connectivity to capture the demand from AI cloud computing companies for computing power sales and leasing, as well as the large-scale deployment of actual AI applications for enterprises.

Last week, two new cloud powerhouses in the US stock market—CoreWeave and Nebius—experienced significant stock price surges after being upgraded by JPMorgan and BNP Paribas, reflecting the market’s sustained focus on computing infrastructure service providers; Globe’s data center joint venture mainly participates in the data center infrastructure link, and its value realization depends directly on customer signings, power system stability, delivery of AI computing capacity, and profitability after leasing.

This strategy already has operational support. Globe disclosed in August this year that the first floor of Fairview Phase I data center has already been contracted by clients, while subsequent floors are under commissioning and development, supporting high-density AI deployment and liquid cooling demands; The joint venture also signed a ten-year renewable energy supply agreement with MPower, involving a total power supply of 40.5 megawatts for the Fairview and Cavite facilities. These developments correspond respectively to client demand and energy procurement, but future returns still depend on construction progress, power costs, utilization rates, and financing terms. For the Philippines, regional capacity spillover creates opportunity, but undersea fiber optic cables, land, GPU/ASIC supply scale, power supply, and compliance capabilities determine whether orders can truly be retained. AI hype provides growth expectations, but delivering on schedule and generating rental cash flow is the key to realizing valuations.

From Connecting Users to Integrating AI Computing Resources: Globe Pursues the Next Billion-Dollar Business

Globe Telecom is betting on the artificial intelligence boom, aiming to develop its data center joint venture into the next billion-dollar business; at the same time, its fintech affiliate Mynt is moving toward what could become a record-setting IPO in the Philippines.

Globe President and CEO Ernest Raymond Cu said in an interview on Monday that the company plans to invest at least $1 billion by 2027, mainly for expansion of its data business. Over the next few years, capital expenditure is expected to account for 23% to 28% of revenue.

“We need to continue investing in the AI super network,” he said in the interview, “and we will continue to do so, but will be very prudent in selecting investment areas, clarifying priorities, focusing mainly on data and digital economy-related areas.”

As AI fuels a global race for computing capacity, the Philippines’ second-largest telecom operator is looking beyond traditional businesses for its next phase of growth. Cloud service providers and technology companies seeking to meet surging demand are channeling billions of dollars into this industry across Asia.

Globe has established a foothold in this area through ST Telemedia Global Data Centres Philippines, a joint venture formed with Globe, its parent Ayala Corporation, and Singapore Technologies Telemedia. According to Cu, this data center operator runs five facilities, with capacity expected to reach 30 to 35 megawatts by the end of 2026, and to expand to 124 megawatts over the following three years through new builds and expanding existing sites.

He said that taking this joint venture public is one of the financing options under consideration for further expansion. “The market gives any business related to the digital economy a pretty high valuation.”

As shown in the figure above, Globe places its data business at the center of its massive capital expenditure plan—data-related projects account for the main portion of capital spending. Source: Globe Telecom 2026H1 Analyst Briefing.

He noted that capacity constraints in Thailand, Singapore, and Malaysia mean that the Philippines will become the next destination for data center clients. “In the next few years, with the development of AI, you could see a three- to five-fold increase in data traffic.”

However, a report from BMI, part of Fitch Solutions, points out that the Philippines cannot rely solely on spillover demand from Malaysia and Indonesia. To compete for large cloud computing projects, the Philippines needs to leverage its advantage in undersea fiber optic cables connecting developed Asian markets—while ensuring steady power and land supply.

The Philippines has 44 data centers, ranking second-to-last in the Asian market tracked by BMI, with around 140 megawatts of operational capacity and another 264 megawatts under construction or in planning.

Domestically, Globe is striving to catch up in a market dominated by competitor PLDT. PLDT’s data center subsidiary VITRO seeks to raise up to PHP 24.2 billion (about $387 million) through a real estate investment trust IPO. Converge ICT Solutions is also expanding its business footprint.

The Philippine government is seeking to attract more data center investment into the country. In July, President Ferdinand Marcos Jr. signed an executive order requiring that the government’s most sensitive digital information must always be under Philippine jurisdiction, which could boost local facility demand.

Before betting on data centers, Globe already had a successful subsidiary in Mynt. Mynt, the parent company of the Philippines’ largest mobile wallet GCash, was valued at $5 billion in its last financing round in 2024. Its IPO is expected to raise about PHP 60 billion next month, potentially becoming the Philippines’ largest ever IPO.

Cu said that as Mynt expands its credit, investment, savings, and insurance offerings, Globe plans to continue as its single largest shareholder.

Strong Demand Linked to AI Applications Must Ultimately Translate to Data Center Cash Flow

Globe’s data center joint venture mainly participates in the data center infrastructure link, and its value realization depends directly on customer signings, 24/7 power supply stability, AI computing capacity delivery, and post-leasing profitability. The strong demand associated with AI applications must ultimately be realized as data center cash flow to reflect the robust growth value of data center operators.

AI agents evolving from answering questions to continuously executing tasks are extending growth opportunities from chip and cloud service providers to large-scale AI data center operators offering comprehensive AI computing infrastructure, power resources, cooling, and high-speed networking systems.

Muse and Astra are poised to further expand the scope of AI agent usage, and the resulting surge in AI inference workloads will simultaneously increase demands for model inference, tool execution, and state management, while intensifying competition for some software companies. GPUs and TPUs handle model computation; CPUs operate browsers, virtual machines, product retrieval, database queries and transaction coordination; HBM and server DRAM support model data, context, and concurrent work environments; enterprise SSDs store product indexes, task records, and persistent states, while high-speed networks and optical interconnects support distributed data exchange.

Muse and the released GPT-6 Astra represent AI agents that are expanding the range of real-world tasks that AI can take on. Their core underlying change is that a user request can trigger multiple rounds of inference, browser operations, code execution and result validation—driving infrastructure demand from model computation to task execution and state preservation. Meta’s technical documentation clearly states that Muse provides users with independent cloud-based virtual machines equipped with CPUs, memory and storage, and supports parallel sub-agent and scheduled tasks. Therefore, GPUs undertake neural network computation, CPUs handle tool execution and task scheduling, while memory and storage retain runtime state, files and long-term memory; when active users, task frequency, and concurrency scale up, demand can rapidly and massively transmit to servers, networks, and data center capacity.