Powering Discovery, Sharing Success, and Building Community
Dear Name,
Welcome to April’s newsletter, highlighting Pawsey’s continued engagement across research, policy, and global collaboration.
Over the past few weeks, our team has been actively contributing to national and international conversations on high‑performance computing, data, AI, and quantum technologies, while strengthening partnerships that support Australia’s research future.
A key highlight was the launch of PULSE Collaborations. Formerly Pawsey Uptake Projects, the refreshed program reinforces close collaboration between researchers and Pawsey staff, supporting skill development, shared best practice, and strengthened research outcomes across disciplines.
We hosted delegates from the Black Swan Summit for tours and discussions on how our energy‑efficient supercomputer works and the integration of quantum and classical systems.
Nationally, we attended Science Meets Parliament – run by Science & Technology Australia – in Canberra. As part of his time in Canberra, our CEO Mark Stickells had the opportunity to join National Computational Infrastructure (NCI) in hosting the Hon Julian Hill MP at NCI – reinforcing the importance of coordinated supercomputing infrastructure to Australia’s research ambitions.
We had visits from ARDC, WEHI and AARNet, and continued engagement with Western Australian universities and research institutes.
PULSE Collaborations (formerly Pawsey Uptake Projects) help Australian-based researchers maximise the impact of their work by teaming up with Pawsey experts.
Open to researchers based at Australian universities, government agencies and research institutions, teams will receive up to 0.20 FTE of dedicated Pawsey staff support over six months.
This means access to benchmarking and code optimisation, GPU acceleration, workflow improvements, advanced visualisation, and our hybrid quantum-classical approaches.
NCI and Pawsey Supplement Allocations for NCMAS 2026
Following the recent National Computational Merit Allocation Scheme (NCMAS) call for applications, NCI and Pawsey Supercomputing Research Centre have worked together to provide a limited, one-off extension of support to a small number of highly ranked applications that were not successful in the 2026 call.
Given the strong demand and overall quality of submissions, a number of meritorious projects narrowly missed out on allocation. To help maintain research continuity, both facilities contributed a modest, jointly coordinated supplement to selected projects.
Our goal was to ensure that affected research groups can maintain continuity of their computational programs, avoiding disruptions that could compromise ongoing work.
Across NCI and Pawsey, 64 projects received this additional support, including projects requesting time on both systems. At NCI, 53 projects were allocated supplementary compute, while Pawsey supported 28 projects. In total, this extension represents a combined contribution of 60,750 KSU – including 39,750 KSU from NCI and 21,000 KSU from Pawsey.
This measure is specific to this allocation round and reflects a shared effort to support the research community during a particularly competitive cycle.
NCI and Pawsey remain committed to working alongside our partners to enable high-quality research through Australia’s national computational infrastructures.
Guest speakers shared their experiences using cloud infrastructure like Nectar, and how it helps accelerate science and offer researchers more computing capabilities.
Our partnership with another NCRIS facility continues to help us offer scientists and researchers the tools they need to answer the next big questions.
Australian BioCommons: Help shape the future of AI in life sciences research and training
29 April 2026 – Online
By sharing the current tools, processes, bottlenecks and skill gaps you experience, you can help prioritise national investments in digital infrastructure and directly shape our upcoming training programs.
What happens when the rapid rise of AI collides with a global memory shortage?
In this ABC Radio interview, NCI Director Andrew Rohl explains how the surge in AI demand is driving unprecedented pressure on high-bandwidth memory, leading to sharp price increases and global supply constraints.
This, in turn, is making it increasingly difficult to upgrade or replace critical supercomputing infrastructure, potentially limiting scientific capability across areas like climate modelling and advanced research.
Professor Rohl is joined by other experts (Caroline McDaid, Work Ventures Dr Sue Keay, UNSW) in the discussion, highlighting the broader global challenge and its implications for research systems worldwide.
ACCESS-ESM1.6: Australia’s Climate Model Takes Centre Stage in Global Climate Research
Australia’s latest Earth system model, ACCESS-ESM1.6, has joined CMIP7, a major international effort shaping future IPCC climate assessments.
Enabled by NCI’s supercomputer, Gadi, the model enables researchers to simulate complex interactions across the atmosphere, oceans and land, generating the data needed to better understand climate change.
This milestone ensures Australian climate science is benchmarked globally while capturing the unique dynamics of our region, supporting more informed decisions around infrastructure, energy and climate resilience.
To celebrate next week's World Quantum Day we will feature a case study to showcase the potential of quantum computing enabled through our Setonix-Q Pilot program.
With something as complex as analysing medical datasets, the team of researchers have been looking at quantum computing to help predict patient outcomes from medical data.
A feature article from Quantum Zeitgeist, called 'Quantum Computing Yields Comparable Accuracy with Six Models', shared early outcomes on a project lead by the Pawsey team with researchers from The University of Western Australia, in collaboration with QuEra Computing.
Using a technique called quantum reservoir computing (QRC) –- a QML technique that uses a quantum system's natural dynamics to process data – researchers tested six classical machine learning models looking for a potential route to improve diagnostics and personalise treatment in the field of medicine.
While simulated (noise-free) quantum computing performed similarly to classical computing, it tended to overfit – this is when machine learning is accurate on training data, but not on new data.
Running QRC on real quantum hardware – in this case, the Aquila neutral atom Rydberg processor – produced more accurate and stable predictions, because the physical process of quantum computation restructured the data in a way that made the model more generalisable – better at predicting outcomes it had never seen before.
The unavoidable noise and physical constraints of the hardware appeared to act as a natural form of regularisation, preventing models from memorising the training data and improving their ability to generalise to new patients.
