Google DeepMind opens AI research lab in Singapore to advance core AI and regional collaboration
Google DeepMind has announced a new AI research hub in Singapore, expanding its footprint across Asia-Pacific. The lab will drive fundamental research while working with local universities, government, and industry to move ideas into tested, real-world use.
The focus is clear: accelerate breakthroughs in machine learning, responsible AI, and applied science. For researchers, this means more shared infrastructure, stronger partnerships, and faster paths from promising models to field deployment.
Why Singapore
DeepMind cites Singapore's talent density, research infrastructure, and policy support as the draw. The aim is to build with the region, not for it, by embedding with existing institutions and public-sector programs.
"Singapore offers a vibrant ecosystem for AI research and innovation. By building a lab here, we can deepen our collaborations and contribute to advancing AI responsibly across the region," said Koray Kavukcuoglu, DeepMind's Vice President of Research.
Core research priorities
- Reasoning: Advancing models that plan, verify, and improve their own outputs.
- Multimodal learning: Integrating text, vision, audio, and structured data for richer inference.
- Efficiency: Methods that reduce compute, energy, and latency without sacrificing quality.
Early collaborations and applied work
- Science: Teams at A*STAR and the National Neuroscience Institute used AlphaFold to advance Parkinson's research, linking immunology with neurodegeneration and opening doors to earlier diagnosis and targeted therapies. Learn more about AlphaFold.
- Public services: With GovTech, the Cyber Security Agency of Singapore, and IMDA, an AI agent sandbox is live to safely test autonomous solutions for service delivery.
- Multilingual AI: With AI Singapore, Project Aquarium launches as an open data platform for Southeast Asian languages. The partnership supports SEA-LION, including the first multimodal model, SEA-LION v4, built on Gemma 3's multimodal capabilities.
- Education: Students in Singapore receive one year of free access to the Google AI Pro Plan, plus Gemini Academy programs via IMDA's Singapore Digital Office to broaden AI literacy.
- Startups: Through Google for Startups: AI First, DeepMind is backing Singaporean founders using generative AI on economic, societal, and environmental problems.
What this means for scientists and research leaders
- Closer loops from lab to field: Policy, public-sector testbeds, and industry pilots reduce the gap between prototypes and adoption.
- Data and benchmarks for SEA: Language resources and multimodal datasets aligned to Southeast Asian contexts will improve evaluation quality.
- Stronger evaluation culture: Expect more work on safety, reliability, and system-level metrics for agents and multimodal models.
- Talent pipelines: Opportunities for joint appointments, visiting researcher roles, and student programs tied to active deployments.
How to engage
- Align proposals to three pillars: reasoning, multimodal learning, and efficiency. State the metric that matters and the path to a field test.
- Leverage the public-sector sandbox for safe trials of agentic systems with clear guardrails and auditability.
- Contribute to Southeast Asian language resources and evaluation suites. Multilingual robustness is a priority signal.
- For biomedical teams, assess AlphaFold-involved pipelines for target discovery and earlier diagnostic signals in neurodegenerative research.
Keep track of updates
Follow official announcements and technical write-ups from DeepMind's team for calls, datasets, and partnership programs. DeepMind blog
If you're building your team's AI capability stack, explore focused training paths by job function: AI courses by job
Your membership also unlocks:
