ST Telemedia Global Data Centres (STT GDC) urged the government to secure sufficient electricity supply, explore the establishment of small modular reactors, and keep power prices competitive for the artificial intelligence (AI)-driven economy.
"Apart from Indonesia, Thailand is one of the highest-growth markets within our footprint in the Southeast Asian region, which comprises more than 100 facilities across 12 global markets with a total capacity exceeding 2 gigawatts," said Budsarin Pradityont, country head at ST Telemedia Global Data Centres (Thailand), also known as STT GDC Thailand.
Currently, it has three data centres, including STT Bangkok 1 (BKK1), located in Hua Mak. The facility has a capacity of 23 megawatts and is currently about 90% utilised by customers, primarily serving enterprise and hyperscale clients.
It is investing US$240 million in its STT BKK2 data centre, with the first phase expected to be operational this year before reaching 24MW in 2027. Its BKK2 facility is situated on the same campus as BKK1.
BKK2 is already 30-40% committed, largely driven by existing hyperscale and AI customers expanding their operations.
The BKK3 facility, with a capacity of 2MW, is located within the One Bangkok project and is currently 40% utilised.
The BKK3 facility specifically targets enterprise customers, multinational companies, and the banking, financial services, and insurance sectors.
It is considering building its fourth data centre, with a capacity of 50MW, in the Eastern Economic Corridor, which would become its first data centre located outside Bangkok.
The fourth centre will support new potential customers, including "neo cloud service providers", such as those offering inference as a service (IaaS), a cloud delivery model that allows organisations to run AI models via application programming interfaces, eliminating the need to own and maintain expensive hardware.
The other customer group it aims to serve is those offering graphics processing unit (GPU)-as-a-service.
Ms Budsarin said these neo cloud service operators come from China and Taiwan.
The fourth data centre requires at least 50MW of capacity, so it needs to ensure a sufficient electricity supply, including renewable energy, she added.
POWERING AI GROWTH
Ms Budsarin said the government should prepare a future electricity plan, as demand from data centres is anticipated to surge, requiring 1GW of capacity to serve AI and future workloads over the next 1-2 years.
The government needs to keep electricity prices stable to maintain Thailand's competitiveness in attracting data centre and hyperscale investments.
She is aware of criticism that data centres consume massive amounts of energy and water while generating relatively little employment.
She recommended that the government introduce measures to attract data centre supply chain manufacturers, such as transformer equipment producers, to Thailand, similar to the approach taken with electric vehicles.
Tiwa Petcharat, the company's executive vice-president for data centre operations, added that data centres employ electrical and mechanical engineers, and that companies should prioritise hiring Thai nationals to create jobs for the local workforce.
He said the immense power required by AI -- particularly GPU clusters -- makes sustainability a critical issue.
Currently, STT GDC purchases power from the Metropolitan Electricity Authority.
Because direct purchasing of pure green energy is limited in the Thai market, STT GDC uses interim solutions such as International Renewable Energy Certificates (I-RECs) to offer "green" options to clients.
STT GDC is actively researching small modular reactors -- a type of advanced nuclear technology-- through a centre of excellence in Singapore.
However, adopting small modular reactors in Thailand will require addressing significant challenges, including public acceptance of nuclear energy, safety concerns, and establishing appropriate government regulations.
AI READINESS GAP
Ms Budsarin also shared a report entitled Mind the Gap: Bridging the AI Infrastructure Readiness Divide, which surveyed 600 organisations in nine Asian countries, including 60 in Thailand.
The survey categorises AI readiness into four stages: explorer, conducting proof of concept but not fully implementing AI; builder, using AI in actual work but in isolated, siloed functions; integrator, integrating AI across various functions and recognising it as a core component of the organisation; and leader/pioneer, using AI as the primary engine for business growth.
The survey found that 78% of Thai companies are at the "builder" stage, higher than the broader Asian average of 71%.
However, Thailand lags in advanced adoption, with only 8% of organisations reaching the "integrator" stage.
Despite 48% of companies being ready to support future AI investments, few are highly successful in driving AI initiatives. Key roadblocks include difficulty measuring return on investment, AI skill shortages, data security concerns, and infrastructure limitations.
To overcome these barriers, businesses can partner with data centre and AI service providers to fill skill gaps, speed up deployment, shift capital expenditure to operating expenditure, and leverage established security standards.
Companies should also broaden how they measure return on investment by considering efficiency gains and customer benefits, while choosing an AI architecture model that aligns with their long-term business and security needs.
