Maybank Investment Bank has foreseen artificial intelligence (AI) demand to spur neocloud race in ASEAN.

The research house says in a note on last Friday that demand for neocloud capacity has gone truly global and hyperscale, exemplified by Microsoft’s multi-year Nebius commitment and Oracle’s bullish cloud guidance.

It is noted that Microsoft’s Nebius deal ($17.4 billion, expandable to $19.4 billion) and Nebius’s immediate $3 billion financing highlight how hyperscalers are contracting third-party GPU capacity to accelerate AI rollouts rather than waiting to build in-house.

Oracle’s upgraded outlook — driven by OCI expansion, hyper-scaler partnerships, and neocloud-style initiatives —underscores enterprise AI demand is both large and accelerating, contributing to a broad re-rating of the AI infrastructure sector and boosting sentiment for specialist GPU providers such as CoreWeave.

Maybank also sees ASEAN telcos poised to capitalize on neocloud opportunity.

Cited ABI Research, it noted the telco industry is set to generate over $21 billion in GPUaaS revenue by 2030, driven by increasing enterprise demand and government requirements for AI sovereignty.

“Within ASEAN telcos, Singtel and Indosat are emerging as regional neocloud players, with Singtel’s RE:AI platform moving from experimentation to scale, while Indosat signals about 3 percent earnings before interest, taxes, depreciation, and amortization (EBITDA) contribution from AI cloud as early as 2026,” said the research house.

It is noted that Singtel is accelerating the scaling up of RE:AI, its sovereign AI cloud platform, by leveraging AI-ready data centers, dense networks, and its Paragon orchestration stack.

The service has advanced from H100 pilots to full GB200 deployment with strong customer uptake, and it plans to be among the first to adopt GB300s.

The business model is shifting towards multi-year, customer-underwritten contracts across government, research, pharmaceutical and enterprise verticals, reducing execution
risk.

Expansion beyond Singapore is underway, with a Japan entry through Hitachi and the formation of a global GPUaaS alliance/partnerships to federate capacity.

While infrastructure assets (such as liquid-cooled AI data centers, cable landing stations and low-latency networks) differentiate the platform, the core opportunity lies in scaling RE:AI as a cost-competitive sovereign alternative to hyperscalers.

Meanwhile, Indosat’s 2Q25 update signals a decisive shift in its AI cloud journey from pilot to scale.

The company has deployed GB200 Blackwell chips and has already onboarded 20-plus customers onto its sovereign AI cloud.

Management guided for $35 million of net new AI revenue in FY25, weighted to the second half of 2025, with about 60 percent EBITDA margin and strong visibility into FY26, underpinned by multi-year contracts.

Strategic partnerships (Kominfo, NVIDIA and Cisco) and the launch of an Indonesia AI Centre of Excellence reinforce Indosat’s position as a national AI enabler.

While capital expenditure (capex) is being frontloaded to accelerate GPU infrastructure build-out, management highlighted AI as a meaningful long-term growth driver.

#Neocloud #AIInfrastructure #GPUaaS #SovereignAI #ASEANTelcos