Open architecture offers 'third way' for Singapore to get ahead of space data centre race
The small state can leverage its neutrality and R&D leadership to champion an open, layered architecture where no single stakeholder owns the whole stack, says Nanyang Technological University (NTU)'s Prof Wen Yonggang.
-1788483683065.jpg)
Currently, the space data centres sector is split into two camps. Singapore could leverage its neutrality and R&D leadership to champion an open, layered architecture where no single stakeholder needs to own the whole stack. Image: Canva
Unlike the US or China, Singapore doesn't have the land, water or power to build data centres at a similar scale.
But there is a "third way" for the small state to influence the space-based data centres sector, according to Nanyang Technological University (NTU)'s Professor and President's Chair of the College of Computing and Data Science, Wen Yonggang.
Instead of looking to build the most infrastructure, Singapore can set an open, layered architecture for other countries and companies to plug into and build on.
He was speaking in the media briefing for the upcoming Singapore Space Summit organised by the National Singapore Space Agency (NSAS).
Currently, the space data centres sector is split into two camps, he said, with the US pursuing it largely commercially, and China pursuing them through state investment. But both tend toward closed systems.
Prof Wen believed that Singapore could leverage its neutrality and R&D leadership to champion an open, layered architecture where no single stakeholder needs to own the whole stack.
This would also give Singapore the opportunity to put forth its own strengths in specific technology layers, like its deeptech advantages.
Building an open architecture
Prof Wen hopes to use Singapore's reputation for neutrality to champion an open, neutral alternative.
"I want to promote an open solution, open architecture, and open reference," he said, calling it "the third way of implementing the space mission."
In practice, that means designing space data centres the way computer networks are designed: as a layered structure, where each layer operates semi-independently, he explained.
A country or company with the best solution, idea or technology for a given layer can plug in, without needing to control the whole stack.
As Singapore has a few decades of R&D to draw from, Prof Wen believed that some of that homegrown tech could prove themselves within an open architecture.
This isn't a new playbook for Singapore, as he pointed out the aviation, maritime and semiconductor industries all followed a similar path with Singapore positioning itself within each global supply chain.
The Singapore advantage
Demonstrating Singapore's thought leadership in space data centres, Prof Wen shared that his team at NTU began researching on the topic back in 2021.
This was years before the idea started to become mainstream when Elon Musk and Jeff Bezos started to announce their own orbital data centre plans.
By this year, one of their papers published in Nature Electronics had become influential enough to be cited by the US Government Accountability Office (U.S. GAO) as a source in a Congress briefing on space data centres, he shared.
He added that the paper later caught the attention of international news outlets like Wall Street Journal.
A robust research and education system invested in training talents and research in emerging tech also plays an important role in building space data centres, he said.
Why Singapore should pursue a space data centre strategy
Prof Wen pointed out the reasons for Singapore to pursue a strategy in space data centres.
Firstly, it is infrastructure that supports Singapore's aspirations to become a global artificial intelligence (AI) powerhouse. The small state is not limited by its resource and is not dependent on other countries for this infrastructure support, he said.
Secondly, space technologies are increasingly dual use given their complex supply chain, he said.
In other words, the use of the technologies is increasingly moving from defence to more widely-used commercial applications.
Prof Wen also pointed to a few concrete milestones that signal Singapore should continue to develop space data centres.
First is with increased power for AI models. Orbital compute nodes, which are high-performance computing satellites, can produce around 10 kilowatts (kW) of power to run a small AI model, enabling the scaling of AI workloads.
He called this a “real breakthrough” as traditional satellites provide less than 5 kW of power, which is enough for only a few small GPUs.
Finally, it is the ability to connect satellites in the same orbit to communicate with one another and offload compute for future 6G satellites.
Demonstrating this 6G offload use case could create real commercial demand for space data centres, beyond just processing satellite imagery data, he shared.