Asean companies keen on AI for business benefits, despite its energy costs

[SINGAPORE] Almost four in five South-east Asian businesses expect the benefits from artificial intelligence to outweigh its energy burden – even as they brace for rising costs and infrastructure pressure.

Out of 514 regional business leaders polled by The Business Times and market research firm Kantar, 78 per cent see AI’s benefits outweighing its energy requirements.

Released on Monday (Oct 12), The Business Times Insights: Powering Asean’s Future 2026 – the second regional survey by BT and Kantar – showed Singapore, Vietnam and Thailand as having the highest levels of optimism.

This comes as AI adoption is accelerating across South-east Asia; 62 per cent of respondents have increased their investments in AI or data-driven technologies over the past 12 months.

Growth is particularly pronounced in Indonesia, where 85 per cent of businesses reported moderate-to-significant increases in AI investment over the last year.

The regional optimism around AI is “no surprise” to Sunil Veetil, whose Singapore advisory Perdura helps companies with green finance and the energy transition.

He noted that AI returns materialise quickly when applied to the right use cases. For instance, it can enable multilingual customer service, providing distinct advantages in a region as diverse as Asean.

In addition, said Manik Bhandari, Asean data and AI leader at EY, energy costs are currently “a very small fraction” of the overall AI token pricing, and thus “quite minimal” when building AI applications in an enterprise setting.

Bracing for costs, infra pressure

That said, there is awareness of energy pressures; 41 per cent of Asean businesses expect a moderate increase in energy demand as AI adoption scales, while another 40 per cent are bracing for a significant increase.

Businesses expect energy demand to manifest in operational challenges:

• Rising electricity or operating costs, as cited by 46 per cent of respondents;
• Pressure to upgrade existing infrastructure (39 per cent); and
• Increased competition for energy resources (37 per cent).

While businesses may not currently be bearing the energy costs of AI, Veetil noted that this “changes when AI becomes captive or domestic, when companies run their own models or countries build sovereign (models) on local power”.

“Then, the energy bill lands on the business.”

Some markets are less optimistic about AI; in the Philippines, just 64 per cent of respondents see AI’s benefits outweighing energy requirements, below the regional average.

One explanation could be that “the trade-off looks different in markets where electricity costs, grid reliability and energy security are already significant business concerns”, said Zayana Zaikariah of the Institute of Strategic and International Studies in Malaysia.

Climate exposure may be another reason.

“In countries facing frequent typhoons, flooding and other disruptions, businesses may place a higher premium on the resilience of electricity and digital infrastructure before adding new sources of demand,” she said.

Beating AI Fomo and fatigue

It is thus critical for businesses to use AI judiciously.

Professor Lawrence Loh of the National University of Singapore (NUS) said that Fomo – a “fear of missing out” – is leading many businesses to step up AI adoption, and that staff and management “may be experiencing severe AI fatigue now”.

“The wrong metric for business is how much AI is being used,” said Prof Loh, who is director of the university’s Centre for Governance and Sustainability. “The right one should be how much measurable economic and other benefits are being derived.”

Bhandari of EY said that before building an AI application, companies should assess not only the token prices, but also the application’s carbon footprint.

Businesses can use smaller, task-specific models that reduce memory usage and can be run on standard hardware or edge devices, with a vastly reduced carbon footprint. “Not only is it cheaper for me as an enterprise, it’s also better for the environment,” he said.

Zayana echoed that businesses should “match the level of compute to the value of the task”.

A compute-intensive AI system “should not be the default if a smaller, task-specific model can achieve the same business outcome”, she said.

Country-level implications

AI-driven hunger for energy has implications not just for businesses, but also for countries’ energy resilience and decarbonisation goals.

If not managed well, greater energy demand from AI could translate to more dependence on fossil fuels, said Dr Dinita Setyawati, external affairs and strategic impact lead for Asia at energy think tank Ember.

Conversely, AI facilities “can actually become part of the solution if they help mobilise investment in renewable energy, storage and the grid infrastructure needed to support that demand”, she noted.

Governments can also play a role with clean power standards for data centres, as well as engaging communities in decisions about where AI facilities are located and how they use energy and water, Dr Dinita said.