Singapore is planning a new western island that could reshape its industrial and energy infrastructure for the next half-century. At the National Day Rally on August 23, Prime Minister Lawrence Wong said the government intends to merge Semakau, Bukom, Pulau Sudong and several smaller islands south of Jurong Island into one larger island over the coming decades.
The project could accommodate advanced manufacturing, power generation and military facilities. Wong did not identify it as an AI or data center development, but its significance for the digital economy is clear. Singapore’s ability to attract large-scale computing infrastructure increasingly depends on whether it can provide enough electricity at a competitive price.
AI infrastructure is often measured by processors, models and data. Its expansion, however, is increasingly determined by power generation, grid access and construction schedules. Advanced chips cannot operate without hundreds of megawatts of reliable electricity, making energy planning inseparable from digital industrial policy.
Singapore’s Energy Market Authority expects peak electricity demand to rise by between 2.4% and 4.8% annually over the next decade, driven partly by semiconductors and data centers. Demand could reach between 9.6 and 11.4 gigawatts by 2031.
The government has invited proposals for at least one 600-megawatt hydrogen-ready gas turbine unit to begin operating in 2031 and up to two more in 2032. Another 600-megawatt unit is scheduled for 2029. By comparison, Singapore’s second Data Centre Call for Application, launched in December 2025, made at least 200 megawatts of new data center capacity available. The figures show how much generation must be secured before additional computing capacity can be approved.
Electricity supply is only part of the challenge. Singapore generates most of its power from imported natural gas, exposing costs to global fuel prices and geopolitical disruption. It is expanding solar capacity toward a target of three gigawatt-peak by 2030, but domestic solar power is expected to meet only about 10% of projected energy needs by 2050.
Efficiency must therefore accompany new generation. Singapore introduced an energy-efficiency standard for data center IT equipment in 2025, targeting savings of at least 30% through more efficient hardware, higher operating temperatures, workload consolidation and better equipment utilization. In a land- and energy-constrained economy, every megawatt saved reduces the amount of new infrastructure required.
A 50-Year Plan Meets a Quarterly AI Cycle
The central tension in the western island project is time. Land reclamation, island integration, power plants and transport links will take decades, while AI processors, cooling technologies and computing architectures change within quarters. The island cannot resolve Singapore’s immediate data center constraints.
Its strategic value instead lies in preserving long-term flexibility. The site could accommodate power generation, energy storage, semiconductors, advanced manufacturing and other industries as technologies and economic priorities evolve. Rather than betting the project on a single AI cycle, Singapore is creating industrial space that can remain useful even when forecasts prove wrong.
The approach follows the logic of Jurong Island. Singapore merged several smaller islands to create an integrated energy and chemicals hub that supported refining, petrochemicals and trade for decades. The proposed western island seeks to provide a similar platform for an era in which electricity and computing could become as strategically important as oil was to the previous industrial economy.
Until the island is developed, Singapore will need stronger regional integration. The Johor–Singapore Special Economic Zone allows companies to combine Singapore’s headquarters, financing, research and intellectual property capabilities with Johor’s land, workforce and industrial resources.
This “twinning” model could extend to digital infrastructure. High-value business functions and customer relationships could remain in Singapore, while some power- and land-intensive computing facilities are distributed across Johor and other parts of Southeast Asia. Singapore would not need to host every server to remain a regional center for financing, governing and securing AI infrastructure.
Cross-border computing nevertheless requires reliable telecommunications, coordinated power development and clear rules for cybersecurity, data movement and service continuity. Moving data centers across a border does not solve the problem if electricity, water and transmission constraints simply reappear in the receiving region.
The Korea Signal
Singapore’s plan carries a direct lesson for South Korea. Korea has leading semiconductor companies and a strong digital economy, but its AI ambitions face the same questions over generation, transmission, electricity prices and regional concentration. GPU purchases and data center announcements do not create computing capacity unless power can be delivered on schedule.
AI policy must therefore connect semiconductors and data centers with power plants, transmission networks, industrial land and cooling systems. Governments should assess AI infrastructure not only by investment value or processor count, but also by the source, cost and delivery date of the electricity it requires.
Singapore’s western island is not an AI island. Its importance lies in recognizing that technological leadership depends on physical infrastructure. The next phase of the AI race will be shaped not only by who designs the best chips and models, but by who can supply the power to keep them running.
