Aspire 2B has more computing power than 120,000 high-end artificial intelligence laptops combined.

SINGAPORE – Research into weather forecasting, disease prediction and chatbots that understand the nuances of Asian languages has received a significant boost with the launch of Singapore’s latest national research supercomputer. Called Aspire 2B, the supercomputer has the country’s largest cluster of advanced graphics processing units (GPUs) from US chipmaker Nvidia, allowing it to have more computing power than 120,000 high-end artificial intelligence laptops combined. This massive computing power will, in turn, help the more than 9,000 public researchers from universities, research institutes and government agencies here build more complex AI models and shorten experimentation time. “We see AI as a multiplier for our research community to redesign your workflows and speed up breakthrough discoveries and applications,” said Minister for Digital Development and Information Josephine Teo during the launch event on June 8 at the Nanyang Technological University (NTU). “With Aspire 2B, models that were previously too large can now be trained in Singapore to meet our specific needs... Workloads that had to be sent overseas can now use our national research infrastructure,” she said. These workloads include advanced climate modelling that combine AI and physics-based simulations, and higher-resolution weather forecasting modelling. “This can help us anticipate intense rainfall and rising seas earlier, and plan our urban development and coastal defences around them,” said Mrs Teo. Aspire 2B, located within NTU’s Innovation Centre, is part of national AI infrastructure that the National Research Foundation (NRF) committed to build in 2024 with a budget of $270 million. The supercomputer is overseen by the National Supercomputing Centre, which was established in 2015 to support the country’s high-performance computing needs. GPUs have become the foundational bedrock of the AI era with its ability to perform technical calculations quickly and at a greater efficiency than central processing units (CPUs), found in most computers. Built with more than 1,500 of Nvidia’s advanced H200 GPUs, Aspire 2B has four times the compute power of its predecessors Aspire 2A and 2A+ combined. The Aspire 2A and 2A+ , which were launched in 2024, has supported over 1,500 projects since 2024. These projects include A*Star’s multilingual AI model Meralion, which can understand several regional languages including Hokkien, Mandarin, Tamil and Malay. Meralion, which can interprete regional accents, dialects and colloquialisms, is helping social service agency Lion Befrienders automate routine check-in calls on seniors. Researchers will be building on this foundation to train larger and more advanced multimodal models to cover more undeserved languages, interpret images, speak more naturally and carry out tasks independently. The Aspire 2A has also supported the development of coastal protection strategies, climate resilience and regional food security under the National Environment Agency’s Third National Climate Change Study. Similarly, local maritime engineering specialist Mencast has used Aspire 2A to shorten the development cycles of marine propellers, producing over 10,000 design iterations within days instead of months. Singapore Medical Foundation AI Model, a national programme launched in 2025 to drive the development of healthcare AI models , will also be using Aspire 2B to train larger models with more diverse health records and clinical datasets in text, photos and audio. In the long run, this will allow hospitals to identify patient health risks much earlier and provide more personalised preventive care, said NSCC. Part of the $270 million NRF commitment has also been spent on equipping 1,000 specialists with the know-how to operate supercomputers, said NSCC’s chief executive Dr Terence Hung. The centre will continue to strengthen the local talent pipeline through structured training, onboarding programmes and technical support for users, which includes workshops, user consultations and hands-on opportunities for researchers and students, said NSCC. Next-generation AI systems such as AI agents and physical AI will require significantly more computational intensity than systems today, and the NSCC must be prepared to support such requests, said Mrs Teo. She added: “Our measure of success cannot be limited to the scale of hardware available – what matters is how well we use infrastructure; use the capacity efficiently, effectively and in service of the research community. “People and skills will still be essential to turn hardware accessibility into robust climate models, medicines, or better tools for industry.”

Physicists at the University of Oxford show they can assemble Schrödinger’s cat-like superpositions from far more exotic quantum ingredients.

Selection marks the next phase of construction on Micron’s New York project, expected to generate 50,000 jobs.

OneAdvanced has launched what it bills as the UK’s first private, sovereign healthcare large language model trained on NHS primary-care data.