Singapore Fusion Power Estimates Span Two Decades as Data Centre Load Grows
Analysts put commercial fusion in Singapore anywhere between the 2030s and 2050, while the city-state already carries 54% of the region's data centre load.
Estimates for when fusion energy could contribute to Singapore's grid span from the 2030s to 2050, a range that reflects genuine disagreement about how quickly the technology can move from demonstration to commercial operation, asian-power.com reported on Monday (2026-09-07).7
Singapore currently carries 1.4 GW of data centre load, amounting to 54% of Southeast Asia's total, according to Wood Mackenzie. That concentration means any gap between when firm low-carbon supply is needed and when new generation actually arrives falls hardest on the city-state.2
Wood Mackenzie projects data centre power demand across Southeast Asia will quadruple from 2.6 GW to 10.7 GW between 2025 and 2035, rising from 1% to 3-4% of regional peak demand. Under a high-demand scenario, where 75% of proposed projects are built rather than the 50% base case, the figure reaches 13.7 GW by 2035. Data centres could account for 7-10% of all Southeast Asian power demand growth over the decade, or 48-70 TWh.2
The investment required is substantial. Bain & Company and Standard Chartered's 2026 Southeast Asia Green Economy Report estimated that incremental power demand from data centres, electric vehicles and green industrial parks will exceed 100 TWh by 2030, requiring more than US$200 billion in total investment, with more than half directed at data centres.1,3
Three conditions need to align for fusion to become a realistic supply option, according to the asian-power.com report published Monday (2026-09-07). The first is a licensing track separate from fission regulation. The second is off-take certainty: committed buyers before construction begins. The third is local supply chain participation, whether through Singapore-based manufacturers or Singapore serving as a hub for the broader Asian supply chain. Capital and talent availability were also identified as prerequisites.7
None of those conditions are close to being met. Malaysia's experience with conventional nuclear illustrates the sequencing problem: data centres are forecast to consume up to 30% of that country's national power supply by 2030, surging from 8.5 TWh, but nuclear capacity is not expected to arrive before that peak, according to TechTimes reporting from August (2026-08-05). Fusion faces the same gap, stretched further by its longer development horizon.6
Grid infrastructure adds pressure from a different direction. Bain and Standard Chartered found that capital is available across Southeast Asia's green economy transition, with the sector projected to grow from US$290 billion in 2025 to US$430 billion by 2030. But execution gaps are slowing how quickly investment converts into operating capacity, and physical grid buildout has lagged demand growth.5,3
Energy security concerns have sharpened the search for alternatives. Several Southeast Asian countries moved to revisit nuclear power plans for AI data centre supply after the Iran war disrupted regional energy supply chains, the Associated Press reported in March (2026-03-25). JKM, the Asian LNG benchmark, was priced at $24.02/MMBtu in Monday's (2026-09-07) session, already making long-term gas-fired generation commitments expensive for power-sector buyers dependent on seaborne imports.4
Singapore's relative share of regional data centre load will shrink over time even as its absolute load grows. Wood Mackenzie projects Malaysia and Thailand will lead the region by 2035 at 4.5 GW and 2.6 GW respectively, outpacing a 1.9 GW forecast for Singapore as land and power constraints push new capacity elsewhere.2
The span between the optimistic and pessimistic fusion timelines hinges on problems that sit outside the physics. A licensing regime distinct from fission does not yet exist in most jurisdictions. Off-take markets for fusion power, the committed buyer contracts that project financiers require before construction can begin, do not exist either. Until both are in place, the 2030s estimate carries aspirational weight, and the interval between it and 2050 represents a real procurement decision that Singapore's grid planners cannot defer indefinitely.7