In the rapidly evolving landscape of computing and space technology, Nanyang Technological University (NTU) in Singapore is pushing boundaries with innovative research into space-based data centres. This cutting-edge field promises to address the growing demands of artificial intelligence (AI) and high-performance computing by leveraging the unique advantages of low Earth orbit (LEO), such as abundant solar energy and natural cooling from the vacuum of space. A recent collaboration between NTU's College of Computing and Data Science (CCDS) and Singapore-based BC Space marks a significant step forward in turning this vision into reality.

The partnership focuses on developing system-level control designs—algorithms that intelligently manage computation, power distribution, and thermal regulation across satellite constellations. This work builds directly on foundational research conducted at NTU, positioning Singapore as a leader in orbital computing amid national efforts to bolster its space ecosystem.

Foundations: NTU's 2025 Breakthrough in Orbital Data Centres

The seeds of this collaboration were planted in a landmark 2025 study published in Nature Electronics by NTU researchers led by Professor Wen Yonggang, Associate Provost (Graduate Education) and Alibaba-NTU President's Chair in Computer Science and Engineering. Titled 'The development of carbon-neutral data centres in space,' the paper proposed a framework for orbital data centres that harness unlimited solar power and the extreme cold of space (averaging 2.7 Kelvin or -270.45°C) for passive radiative cooling.

Traditional Earth-bound data centres consume vast resources: in Singapore, they account for 7% of national electricity, projected to rise to 12% by 2030 due to AI growth. Space-based alternatives eliminate land scarcity issues—critical for dense urban hubs like Singapore—and offer net-zero emissions potential. The study introduced a life-cycle carbon usage effectiveness (CUE) metric, demonstrating that launch emissions could be offset within years through solar operation.

Two models were outlined: orbital edge data centres, where AI-equipped satellites process raw data (e.g., imaging) on-board, slashing transmission needs by over 100 times; and orbital cloud data centres, constellations of server-laden satellites handling complex tasks like AI training. Simulations via NTU spin-off Red Dot Analytics validated feasibility using space-grade chips from AMD and fault-tolerant tech from Zero Error Systems.

This research not only highlighted advantages but also pinpointed needs for advanced control systems, setting the stage for the BC Space partnership.

The NTU-BC Space Partnership: From Concept to Control

Announced on March 9, 2026, the collaboration between CCDS and BC Space—a Singapore firm pioneering orbital compute infrastructure—transitions from theoretical models to practical system-level control design. BC Space's expertise in neutral orbit AI platforms complements NTU's algorithmic prowess, aiming to operationalize space data centres.

Academic leads Professor Tan Rui and Professor Wen Yonggang bring deep expertise in AI optimization and sustainable computing. On the industry side, Dr. Yang Ze and Dr. Fei Shengkang from BC Space provide hardware access. The partnership addresses core operational challenges unique to space: intermittent solar availability during eclipses, extreme thermal swings (-150°C to 120°C), radiation hardening, and constellation-wide coordination.

By deploying algorithms on real satellites, the project will validate distributed, energy-aware computing systems, fostering Singapore's nascent space R&D ecosystem.

Power Control and Energy Optimisation Algorithms

One pillar of the research is power management. Solar panels in orbit yield 40% more efficiency than ground-based arrays due to no atmospheric interference, but eclipses (up to 36 minutes per orbit) demand smart allocation. Algorithms will predict solar influx, dynamically throttle compute loads, and balance energy across satellites via inter-links.

• Predictive forecasting using AI to preempt power dips.
• Load migration between satellites for redundancy.
• Integration with regenerative fuel cells for backups.

This ensures uninterrupted AI workloads, critical for edge applications like real-time Earth observation.

Cooling and Thermal Management in the Void

Earth data centres guzzle 40% of energy on cooling; space flips this script. Deep space acts as a heat sink, but satellites must radiate infrared via deployable panels without convective air. Challenges include solar heat absorption and internal hotspots from processors.

NTU-BC Space algorithms model thermal dynamics step-by-step: sensor data feeds AI models predicting heat buildup, triggering radiator deployment or workload shifts. Simulations show 90%+ efficiency gains over terrestrial liquid cooling.

Scheduling and Resource Allocation for Constellations

Managing a swarm of satellites requires constellation-scale scheduling. Tasks like AI inference must factor orbital positions, bandwidth, and power budgets. Reinforcement learning algorithms will optimize job queuing, prioritizing low-latency edge tasks while batching cloud jobs during peak solar.

• Dynamic task offloading via laser inter-satellite links.
• Fault tolerance against single-satellite failures.
• Scalability for 100+ satellite fleets.

Early prototypes aim for 50% better throughput than static schedules.

Hands-On Testing: Leveraging BC Space Satellites

BC Space's computational satellites, launched late 2025 and 2026, offer in-orbit validation. Algorithms will run on radiation-hardened hardware, with telemetry beamed to NTU ground stations for refinement. This closed-loop testing accelerates iteration, from simulation to deployment in months.

Success metrics: energy efficiency >95%, thermal stability within 5°C, and zero mission failures due to scheduling.

Singapore's Space Momentum: STDP and SAP Integration

This project aligns with Singapore's Space Technology Development Programme (STDP 2.0) and Space Access Programme (SAP), funding annual launches from 2026-2028 via OSTIn. NTU's concurrent efforts—like AI nanosatellites with Satoro Space and perovskite solar cells—complement orbital computing.

By 2030, STDP aims for commercial space viability, with NTU leading in AI-space fusion. For aspiring researchers, explore research jobs or higher ed jobs in Singapore's booming sector.

Overcoming Key Challenges in Orbital Computing

Despite promise, hurdles persist: launch costs (offset by reusables like SpaceX), radiation flipping bits (mitigated by error-correcting codes), and latency for non-edge tasks. NTU's fault-tolerant designs and BC Space's neutral orbits minimize debris risks.

Solutions include hybrid edge-cloud architectures and AI-driven anomaly detection, ensuring reliability.

Implications for Sustainable AI and Global Computing

Space data centres could cut global data centre emissions (2-3% of electricity) while scaling AI 10x. For Singapore, it secures compute sovereignty amid land limits. Broader: enables climate modeling, disaster response, and precision agriculture via low-latency orbital edge AI.

Stakeholders—from governments to tech giants—eye this; Elon Musk's xAI-SpaceX merger signals industry momentum.

Photo by Galen Crout on Unsplash

Career Opportunities and Next Steps in Space Tech

NTU's CCDS seeks talent in AI, control systems, and space engineering. Check faculty positions, research assistant jobs, or higher ed career advice for pathways. Students, rate professors at Rate My Professor and explore university jobs.

With launches imminent, this collaboration heralds a new era for Singapore's higher education and space innovation.