Singapore captures 99% of Southeast Asia AI infrastructure funding
Deal count hit record high in 2025 as capital remained highly concentrated.
Singapore accounted for about 99% of all disclosed equity funding in Southeast Asia’s artificial intelligence (AI) infrastructure sector, according to Tracxn’s report Building the Stack: Southeast Asia’s AI Infrastructure Ecosystem.
The report estimated about $1.53b (US$1.2b) in total disclosed equity funding between 2019 and 2026 year-to-date, with capital heavily concentrated in Singapore-domiciled companies.
Tracxn said the sector evolved through three phases, namely a formation period from 2019 to 2022, a funding inflection in 2023 to 2024, and a broader but lower-value phase in 2025 to 2026 YTD.
Between 2019 and 2022, the sector recorded 11 funding rounds with about $8.95m (US$7m) in disclosed equity, largely at seed stage.
Funding rose to $324.84m (US$254m) in 2023 and $785.24m (US$614m) in 2024, with the 2024 total including two early-stage transactions accounting for about $777.57m (US$608m).
In 2025, the sector recorded 11 funding rounds, the highest annual deal count in the dataset, but total funding declined to about $409.25m (US$320m).
Seed-stage activity held at seven rounds, whilst early-stage funding fell 48% year on year (YoY).
Year to date, two seed-stage rounds totalling about $1.28m (US$1m) were recorded.
Across 2023 to 2025, early-stage rounds accounted for more than 95% of total disclosed capital whilst no late-stage funding was recorded in the period.
Singapore captured about 99% of total disclosed equity funding across the dataset, whilst Malaysia recorded about US$1.5m. Indonesia and Thailand saw no disclosed deals.
The report noted that the concentration reflects company domicile rather than operational footprint, with regional data centre expansion continuing outside Singapore.
Malaysia, for example, has more than 1GW of operational capacity and a pipeline exceeding 6GW, according to external estimates cited.
Three companies account for about $1.92b (US$1.2b), or roughly 98% of total funding.
MiniMax, founded in 2022, accounts for the bulk of capital and is the only company in the dataset to have reached the public markets following its Hong Kong IPO in January 2026.
Bifrost has raised $17.9m (US$14m) in total equity funding, whilst Aethir has raised about $11.51m (US$9m) at seed stage.
The companies operate across foundation models, training data, and compute infrastructure layers.
The sector recorded two acquisitions in 2025, both involving Singapore-based companies. SUPA was acquired by TDCX, whilst Nidum was acquired by BitsCrunch.
MiniMax’s listing on the Hong Kong Stock Exchange in January 2026 was the largest capital market event in the dataset and the only IPO recorded.
Tracxn said the 2025 cohort of seven seed-stage rounds forms the main pipeline for potential follow-on activity over the next 18 to 24 months, with no late-stage funding recorded outside MiniMax’s public listing trajectory.
($1.28 = US$1)