Singapore expands clinical AI, Nvidia hub

1/ Singapore is putting two pieces in place at once: clinical AI deployment in healthcare and new AI infrastructure partnerships. On May 19-20, officials and partners outlined healthcare programs moving beyond pilots, while Nvidia said it will open a Singapore research hub and support a physical-AI testbed. (techedt.com) 2/ In healthcare, the immediate shift is from research projects to tools closer to clinical use. Tech Edition reported that Singapore used the AI in Health x ATxSummit event on May 19 to announce two MOUs tied to clinical AI development, research translation and regional healthcare collaboration with Bhutan. (techedt.com) 3/ One of those agreements links Singapore General Hospital with A*STAR’s Diagnostics Development Hub. The projects cited include an in-vitro Antibiotic Combination Test, PENSIEVE-AI for early memory problems in seniors, and HealthVector Diabetes. (techedt.com) 4/ The policy backdrop is demographic. Singapore said one in four residents will be 65 or older by 2030, and the government is tying AI adoption to prevention, diagnosis and care delivery rather than treating it as a stand-alone experiment. (techedt.com) 5/ This is not a brand-new healthcare push. Singapore’s Ministry of Health has already said it is applying AI in health institutions, building predictive preventive care and upgrading IT systems to support those capabilities. MOH said it is injecting about S$200 million over five years through its Health Innovation Fund to support development and test-bedding, including AI. (moh.gov.sg) 6/ MOH has also been explicit about where it wants scale first. The ministry said it plans system-wide rollout of generative AI for routine documentation and broader use of imaging AI, including work on breast-cancer screening workflows and the AimSG platform for accessing and monitoring imaging models across public hospitals. (moh.gov.sg) 7/ Governance is central to the rollout. On March 10, 2026, Singapore’s Ministry of Health and Health Sciences Authority published refreshed AI in Healthcare Guidelines, known as AIHGle 2.0, covering accountability, transparency, risk mitigation and categories including both clinical and “clinical-ops” use cases. (bakermckenzie.com) 8/ That matters because Singapore is trying to move AI into regulated settings without separating deployment from oversight. The updated guidelines describe responsibilities for developers, deployers and healthcare professionals, and emphasize safety, fairness, transparency, explainability, robustness, security and data protection. (bakermckenzie.com) 9/ Alongside that healthcare track, Singapore updated its broader national AI agenda on May 20. Smart Nation Singapore said the National AI Strategy was refreshed with 10 priorities, after the National AI Council was established in February 2026 and chaired by Prime Minister Lawrence Wong. (smartnation.gov.sg) 10/ Nvidia fits into that wider push. CNBC reported on May 20 that Nvidia will launch its first Singapore research hub, and only its second such presence in Asia-Pacific. The lab will focus on embodied AI and on improving AI infrastructure efficiency with universities, industry partners and government agencies. (cnbc.com) 11/ Singapore also said it will launch a testbed later in 2026 to help companies co-design, deploy, test and validate commercially viable AI robotic technologies. CNBC said early participants are expected to include Certis, DHL, Grab and QuikBot, with additional trials through a new Center for Intelligent Robotics. (cnbc.com) 12/ The physical-AI piece is mostly about robots, logistics and service operations for now. But it also shows how Singapore wants to position itself: as a place where AI systems are tested in real-world environments, not just trained in labs. That is an inference from the government’s NAIS update and the ATxSummit announcements. (smartnation.gov.sg) 13/ Nvidia’s hardware rollout adds another layer. Nvidia said on May 18 that its first Vera CPU systems were delivered to Anthropic, OpenAI, Oracle Cloud Infrastructure and SpaceXAI, with Ian Buck hand-delivering the systems. The company described Vera as its first custom CPU, built for “agentic AI.” (blogs.nvidia.com) 14/ Nvidia says Vera has 88 custom Olympus cores, 1.2TB/s of memory bandwidth and 50% faster per-core performance. In the company’s framing, the pitch is that agentic systems create heavy CPU demand for tool use, orchestration, retrieval and sandboxed execution, not just GPU demand. (blogs.nvidia.com) 15/ Anthropic’s James Bradbury, quoted in Nvidia’s post, said “Scaling compute is an important accelerant for the growth of models” and called Vera “a promising part of the ecosystem” for agentic workloads. That is Nvidia’s own source, but it is a useful signal about where large AI labs say bottlenecks are moving. (blogs.nvidia.com) 16/ Put together, the Singapore and Nvidia announcements point to a simple story: one side is building governed deployment environments, especially in healthcare and public-sector settings; the other is expanding the compute stack aimed at running more complex AI systems at scale. (techedt.com) 17/ For healthcare specifically, the near-term implication is not that hospitals suddenly jump to autonomous medicine. It is that regulated pilots can move faster when there is already a policy framework for clinical and clinical-ops AI, plus national backing for deployment and test-bedding. That is an inference supported by Singapore’s MOH, HSA and Smart Nation materials. (bakermckenzie.com) 18/ The next things to watch are concrete. Later in 2026, Singapore plans to launch the physical-AI testbed; its broader AI strategy now sits under a refreshed 10-priority agenda; and healthcare deployment will be measured by whether named programs such as imaging AI, documentation tools and the SGH-A*STAR pipeline move into routine use. (cnbc.com)