A*STAR

A*STAR is Singapore's largest public research institution, with 20+ research institutes (I2R, IHPC, SIMTech, IMRE, IBN, CFAR, etc.) covering ICT, biomedicine, chemical materials, manufacturing, sustainable energy and more.

The core units directly relevant to AI:

• I2R (Institute for Infocomm Research): Singapore's largest ICT research institute; the traditional powerhouse for AI / computer vision / NLP / machine learning
• CFAR (Centre for Frontier AI Research): a "frontier AI centre" set up in 2022, focused on LLMs, generative AI, and AI for Science
• IHPC (Institute of High Performance Computing): intersection of supercomputing and AI for Science

A*STAR's positioning is very clear: do the middle-layer research that enterprises won't do and universities can't go deep on. Research outputs are commercialised through ETPL (Exploit Technologies Pte Ltd) and licensed to local enterprises.

A*STAR sits in Singapore's AI ecosystem as the "infrastructure layer" — a lot of what it does goes unattributed.

Specific contributions:

• CFAR co-incubates SEA-LION with AISG: CFAR provides LLM training experience and compute, AISG handles engineering and productisation
• I2R is the "ancestor" of Singapore's NLP / CV research: early SGNLP (later transferred to AISG), Singapore English dialect recognition, and more all originated in I2R
• AI for Science: A*STAR uses AI for discovery in materials, biology and chemistry — one of few places in Singapore with a complete AI4S loop
• Operator of the National Supercomputing Centre (NSCC): SEA-LION and local research compute all depend on A*STAR coordination

On the technical track, A*STAR does not chase "world-best papers" but "industrial deployability" — the fundamental difference from purely academic institutions like NUS / NTU.

A*STAR plays the most easily underestimated yet most indispensable role in Singapore's AI strategy.

In the seven-lever framework:

• Lever 1 (foundational research): A*STAR is the main recipient of national research funding; CFAR benchmarks directly against the global frontier
• Lever 3 (industry adoption): through I2R / SIMTech, transfers AI tech to local manufacturing, healthcare, finance
• Lever 5 (government adoption): many government AI projects (medical imaging, Smart Nation platforms) sit on A*STAR tech stack

A take: **A*STAR's real value lies not in publication output but in its role as a "national tech translator"** — gluing university research, enterprise needs, and national strategy together. It is one of few institutions in Singapore that can do "from foundational research to product deployment" end-to-end.

But A*STAR has obvious bottlenecks: less attractive than universities or big tech (mid-tier salary, narrow promotion path), research efficiency dragged by hierarchical management, and some institutes' research directions are dated (this is a 1991-vintage agency). In the NAIS 2.0 era, A*STAR's role in AI needs to be reassessed — continue as the "applied research middle platform", or focus on a few frontiers (like CFAR)?