The research, led by Ke Zhang of Henan Normal University together with Achyut Shankar of the University of Warwick, Sang-Bing Tsai of the International Engineering and Technology Institute in Hong Kong, and Wattana Viriyasitavat of Chulalongkorn University in Bangkok, addresses what the authors identify as the central obstacle facing contemporary recommender systems: not a shortage of data, but the difficulty of integrating large-scale heterogeneous music-traffic data in a way that maximizes its value. As information overload intensifies and data volumes grow, the sheer computational burden of extracting useful signals from the noise has become as important as the accuracy of the recommendations themselves.

At the heart of the new work is a music recommendation model built on long and short-term preference modeling. The underlying insight is intuitive: a listener’s taste is layered. Some patterns are stable for years, such as a durable affinity for jazz or a favorite era of rock, while others flicker in and out over days or weeks, driven by mood, season, or a single catchy song discovered on a commute. A system that treats all history equally risks drowning durable preferences in transient noise, while one that focuses only on recent behavior loses the deep context that makes long-term recommendations feel personal. The proposed model constructs a user preference model from historical music behaviors, explicitly separating these temporal layers so that recommendations can draw on both the enduring core and the volatile surface of a listener’s habits.

Technically, the modeling leans on machine learning techniques suited to sequential behavioral data, in line with the long short-term modeling tradition that underpins modern sequence-aware recommenders. By learning representations of user interactions that preserve temporal structure, the system can weigh the recency and the persistence of different signals. The authors report that ablation trials, in which components of the model are systematically removed to test their individual contributions, validate the effectiveness of this design, and that the resulting recommendation model outperforms benchmark methods in their experiments.

But accuracy alone is not the paper’s real ambition. Large-scale music-traffic data is plagued by imbalance and sparsity: a small fraction of extremely popular tracks attracts the overwhelming majority of interactions, while the long tail of the catalog is listened to so rarely that the user-item matrix is mostly empty. Traditional collaborative filtering approaches struggle in this regime, producing unreliable estimates for the sparse regions where novelty-seeking listeners actually live. To cope, the study proposes a two-stage decomposition method within non-negative matrix factorization, a technique that factorizes the large user-item interaction matrix into lower-dimensional non-negative components whose additive structure makes them interpretable as latent preferences and latent item attributes.

The two-stage strategy effectively breaks the hard problem into more tractable pieces. Instead of forcing a single factorization to explain both the dense, popularity-dominated region of the matrix and the sparse long tail simultaneously, the method decomposes the problem in stages, which the authors show mitigates the distortion that imbalance otherwise introduces. This matters commercially as much as scientifically: recommender systems that only amplify hits trap users in feedback loops, while systems that can model sparse interactions credibly can surface catalog depth, benefiting artists and listeners alike. The paper frames this as the key to alleviating the data sparsity problems that have long limited recommendation quality on massive heterogeneous platforms.

Perhaps the most distinctive contribution, however, is aimed at a problem that rarely appears in recommendation papers: energy consumption. As the scale of music-traffic data grows, so does the power draw of the server fleets that crunch it. Training and serving recommendation models over billions of interactions is an energy-intensive operation, and the authors argue that achieving low-power processing of these algorithms has become an urgent requirement for energy-efficient data analysis. In response, they design an energy-efficient scheduling strategy specifically for heterogeneous music-traffic workloads, orchestrating computational tasks so that the analytical pipeline consumes less power without sacrificing the quality of the resulting model.

The experimental results reported in the study support both halves of this dual objective. The proposed recommendation model performs better than the benchmarks against which it was tested, and the experiments also verify the effectiveness of the proposed algorithm on energy efficiency, suggesting that accuracy and sustainability need not be traded off against each other. For an industry in which streaming platforms operate some of the largest machine learning deployments in existence, the demonstration that scheduling-aware, energy-conscious design can coexist with improved recommendations is a notable datapoint in a broader conversation about the carbon footprint of artificial intelligence.

The work also reflects a wider shift in how big data research frames its problems. Rather than treating a recommender as an isolated algorithm, the authors treat it as a system embedded in a data pipeline with physical costs: heterogeneous inputs must be integrated, sparse signals must be strengthened, and every matrix operation has an electricity bill attached. Their framework, spanning preference modeling, two-stage matrix decomposition, and energy-aware scheduling, reads as an attempt to close that loop from raw traffic data all the way to sustainable serving. The article was received in October 2023, accepted in September 2026, and published as an open-access paper that is citable under a permanent DOI, with the authors declaring no competing interests and no specific funding support.

For listeners, the practical upshot is subtle but real: better long and short-term preference modeling means the next recommended track is more likely to feel like a genuine reflection of taste rather than an echo of the last three songs played. For operators, the message is louder. As catalogs and user bases expand, the bottleneck is shifting from model accuracy to data integration and energy economics, and methods like those proposed here, which attack sparsity, heterogeneity and power consumption in a single design, offer a template for building recommendation systems that can scale responsibly into an era of ever-bigger music-traffic data.

Non-negative matrix factorization has a long history in recommendation research precisely because of its interpretability. Unlike factorization methods that allow negative values, the non-negativity constraint means latent factors can only be added together, not subtracted, which encourages parts-based representations: a user’s profile becomes a weighted combination of coherent taste components rather than an abstract vector that resists human inspection. The two-stage decomposition proposed in this study builds on that foundation, and the reported ablation trials, a methodology in which individual components are removed one at a time to measure their contribution, offer a level of component-level accountability that single end-to-end accuracy comparisons often lack.

The emphasis on temporal preference modeling also connects to a broader lineage of sequence-aware recommendation. Recurrent architectures in the long short-term memory tradition were designed to preserve information over long input sequences while selectively forgetting irrelevant detail, a property that maps naturally onto listening behavior, where a single skipped track carries different weight than a track played to completion dozens of times. Treating short-term and long-term preferences as distinct modeling targets, rather than collapsing all history into one aggregate profile, reflects a growing consensus that recency and persistence encode different kinds of user intent.

The energy dimension of the work sits within a wider research conversation about the computational cost of machine learning at scale. Large recommendation deployments run continuously rather than in discrete training bursts, meaning that inference and data processing, not just model training, dominate lifetime energy use. Scheduling strategies that route heterogeneous workloads intelligently across computing resources can therefore yield savings that compound over millions of daily recommendation requests, which is why the authors frame low-power processing as an urgent requirement rather than an optimization afterthought.

It is also worth noting the publication trajectory of the paper itself. The manuscript was received in late 2023 and accepted nearly three years later, a timeline that reflects the extended peer review cycles common for work spanning multiple technical domains. It appears as an open-access article under a Creative Commons license that permits non-commercial sharing with attribution, and it is published as a citable, DOI-bearing version ahead of final editorial formatting, an increasingly common practice intended to accelerate access to accepted research.

The collaborative composition of the author team, spanning institutions in China, the United Kingdom, Hong Kong, and Thailand, mirrors the global character of the problem being studied. Music-traffic data crosses borders effortlessly, and the engineering challenges of integrating heterogeneous streams, correcting for sparsity, and constraining energy use are shared by platforms regardless of where their users live. Work that treats these as a single coupled design problem, rather than as separable concerns handed to different teams, offers a useful reference point for how large-scale data systems research may continue to evolve.

Subject of Research: Machine learning-based analysis of large-scale heterogeneous music-traffic data for energy-efficient personalized music recommendation

Article Title: ML-driven large-scale heterogeneous music-traffic data analysis

Article References: Zhang, K., Shankar, A., Tsai, S.-B., & Viriyasitavat, W. (2026). ML-driven large-scale heterogeneous music-traffic data analysis. Journal of Big Data. https://doi.org/10.1186/s40537-026-01557-8

Image Credits: AI Generated

DOI: 10.1186/s40537-026-01557-8

Keywords: music recommendation, machine learning, heterogeneous music-traffic data, non-negative matrix factorization, LSTM, user preference modeling, energy-efficient computing, data sparsity, big data, recommender systems, long short-term preferences, energy-aware scheduling

Among the emerging architectures, in-memory computing has attracted significant attention. This approach eliminates—or at least drastically reduces—the latency and energy overhead associated with shuttling data between processors and memory by performing calculations directly within memory units themselves. Such a paradigm shift promises to revolutionize how complex data-intensive tasks are handled, potentially paving the way for more energy-efficient, faster, and scalable computing solutions.

Building upon these ideas, a pioneering research team led by Professor Xian-Min Jin at Shanghai Jiao Tong University has unveiled a groundbreaking quantum-enhanced in-memory stochastic computing system. Their work, recently published in Light: Science & Applications, represents a milestone in leveraging quantum memory technology at room temperature for practical computational tasks. This system capitalizes on the inherent randomness of quantum mechanical processes, integrating stochastic computing principles with quantum physics to achieve secure and efficient operation.

Central to this innovation is a quantum memory composed of cesium atoms maintained at ambient conditions. This memory harnesses controlled light-matter interactions to generate correlated photon pairs—Stokes and anti-Stokes photons—which serve as fundamental information carriers in the computational process. The unique probabilistic nature of photon emission in this setup enables the encoding and manipulation of data via precisely engineered laser pulse sequences, whose energy and timing modulate atomic excitations within the ensemble.

The computational paradigm revolves around mapping mathematical operations such as addition and multiplication onto the stochastic behavior of photon generation and detection events. Addition is realized straightforwardly by tallying accumulated Stokes photon counts, while multiplication emerges from analyzing the temporal coincidences between correlated Stokes and anti-Stokes photons, reflecting joint event probabilities. This nuanced utilization of quantum correlations imbues the system with natural support for stochastic arithmetic, a feature that distinguishes it from classical counterparts.

Beyond efficiency, security is a pivotal characteristic of the quantum-enabled in-memory computing scheme. Due to the fundamental uncertainty governing photon generation and detection, intercepted partial data fragments reveal no concrete information about the overall computational outcomes. This intrinsic security through randomness constitutes a formidable barrier against eavesdropping, presenting a promising framework for secure remote computation—a growing concern in today’s interconnected digital landscape.

Furthermore, the implementation harnesses quantum correlations to accelerate computational throughput. Remarkably, despite an imperfect retrieval efficiency of just 0.3%, the system demonstrates higher rates of detection coincidences compared to classical stochastic computing methodologies. This advantage not only emphasizes the value of quantum effects in practical tasks but also underscores the potential for improving performance even when hardware imperfections exist.

Professor Jin emphasized the transformational implications of their findings: “Our demonstration reveals that even quantum memories with modest efficiencies can perform meaningful computing operations. This opens up avenues to harness imperfect quantum technologies for real-world applications, broadening the horizon of quantum-enhanced information processing.” Their optimism reflects the broader vision of integrating quantum phenomena into mainstream computational systems.

Looking forward, the research team advocates for the fusion of their quantum memory setup with advanced photonic chip technology and spatial multiplexing schemes. Such integration aims to shrink system footprints while enabling massive parallelism and scalability—critical requirements for deploying usable quantum computing devices outside laboratory settings. The room-temperature operation of their quantum memory aligns with the goal of practical, deployable hardware that circumvents the complexities of cryogenic cooling.

This breakthrough also resonates within the field of quantum-secure communications, as the stochastic in-memory computing platform naturally generates outputs that are resistant to interception or tampering. By leveraging photon statistics and quantum correlations in computational workflows, novel protocols for distributed and remote computing could strengthen data privacy without sacrificing efficiency. Consequently, the approach may inspire innovative solutions that blend computation and security in a unified quantum framework.

Moreover, the work exemplifies how merging concepts from distinct disciplines—quantum optics, atomic physics, and computational theory—can lead to revolutionary architectures. The quantum memory, acting as an information processing resource beyond traditional storage roles, offers a glimpse into future computing landscapes where the boundaries between memory and processor blur. Such hybridization has the potential to redefine algorithmic design and hardware construction paradigms.

In conclusion, the quantum-enhanced reconfigurable in-memory stochastic computing system marks an essential step towards realizing practical quantum technologies embedded within everyday computational devices. Its room-temperature operation, innate stochasticity, and security features collectively present a compelling outlook for next-generation information processing. As research progresses and hardware evolves, we may witness novel quantum computing platforms that integrate seamlessly into data centers, networks, and edge devices—ushering in a new era of computing empowered by quantum physics.

Subject of Research: Quantum-enhanced in-memory stochastic computing with room-temperature quantum memory.

Article Title: Quantum-enhanced Reconfigurable In-memory Stochastic Computing.

Web References: 10.1038/s41377-025-02181-6

Image Credits: Xian-Min Jin et al.

Keywords

Quantum memory, stochastic computing, in-memory computing, cesium atomic ensemble, photon correlation, quantum-enhanced computing, room-temperature quantum device, secure remote computing, quantum optics, light-matter interaction, photonic integration, scalable quantum technology

In traditional computing architectures, particularly the von Neumann model, memory units and processing cores are separate. This separation necessitates constant data movement between the two, leading to significant latency and energy inefficiency. As computing scales up, the limitations of this model become increasingly pronounced. In contrast, neuromorphic photonics enables computation through the use of light, which inherently allows for ultra-high bandwidth and low latency operations. By leveraging the unique properties of light, researchers propose a more efficient way to execute critical computations, such as matrix-vector multiplication, significantly reducing power consumption and increasing operational speeds.

Central to this evolution in computational technology is the concept of integrated photonic neural networks (IPNNs). These networks utilize a trio of foundational building blocks: photonic synapses, photonic neurons, and photonic memristors. Photonic synapses play a crucial role in weight storage and loading, essential for neural network operations. Various devices can be used as photonic synapses, including microring resonators (MRRs), Mach-Zehnder interferometers (MZIs), and phase-change materials (PCMs). Each of these devices serves a specific purpose, maximizing energy efficiency while ensuring compactness and effectiveness in weight storage.

Photonic neurons are equally vital, serving as the nonlinear activation units within the network. By favoring all-optical activation schemes, the researchers highlight the potential for increased efficiency in neural network applications. Furthermore, photonic memristors provide crucial memory capabilities, allowing for both volatile and non-volatile storage. This ability to store and process data at optical speeds presents an advantage over traditional electronic counterparts, thereby facilitating real-time data manipulation.

The review also delves into various architectures that have been developed for integrated photonic neural networks. Among them are coherent networks, which may utilize structures such as MZI meshes to enable rapid on-chip training. Additionally, researchers discuss parallelized IPNNs that employ multiplexing techniques for a significant increase in data throughput. Integrated diffractive networks are identified as a promising architecture for low-latency inference tasks, while reservoir computing emerges as a versatile approach for processing dynamic signals.

Despite the promising advancements in IPNN technology, the researchers emphasize that several hurdles must be addressed for the successful deployment of these systems. Calibration and stability remain critical challenges that need to be navigated. The seamless integration of photonic and electronic components is paramount, particularly in efforts to develop programmable, general-purpose architectures capable of efficient training. Overcoming these obstacles is essential for translating research breakthroughs into practical applications that could revolutionize fields such as edge computing, autonomous driving, and intelligent manufacturing.

The outlook for IPNNs remains positively charged, with the researchers proposing that future developments in optoelectronic integration and programmable platforms will significantly improve robustness and performance. As the team continues to investigate and refine materials like phase-change compounds, microcombs, and advanced multiplexing techniques, the prospect of widespread adoption of photonic neuromorphic computing seems more attainable than ever. This could lead to a paradigm shift in artificial intelligence, transforming how computations are performed and ushering in a new era of energy-efficient, high-speed computing.

To achieve a broader understanding, the authors provide a roadmap that lays out the necessary advancements needed to realize the full potential of photonic neural networks. Breakthroughs in achieving low-energy nonlinearities are highlighted as key objectives, as are initiatives aimed at enhancing storage capabilities, calibration stability for large arrays, and improving photonic-electronic co-packaging. This proactive approach indicates a clear trajectory toward developing photonic AI systems that are not only capable of handling significant computational loads but are also energy-efficient and scalable.

In sum, the future of computing may very well lie in the intersection of photonics and artificial intelligence, as evidenced by the promising developments in integrated photonic neural networks. Researchers anticipate that as foundational technologies and architectures continue to evolve, photonic neuromorphic computing will become a reality, paving the way for intelligent systems that can operate at unprecedented speeds while minimizing energy consumption. This signals a revolutionary step forward, where traditional limits imposed by electronic processing may soon be eclipsed by the versatility and efficiency provided by light-based computation.

Subject of Research: Integrated Photonic Neural Networks
Article Title: Integrated photonic synapses, neurons, memristors, and neural networks for photonic neuromorphic computing
News Publication Date: 2025
Web References: https://www.oejournal.org/oet/archive_list_en
References: Han SF, Shen WH, Gu M, et al. Integrated photonic synapses, neurons, memristors, and neural networks for photonic neuromorphic computing. Opto-Electron Technol 1, 250011 (2025). DOI: 10.29026/oet.2025.250011
Image Credits: OET

The intricacies of matrix operations, particularly in systems that demand high-speed computations for processing vast volumes of data, necessitate innovative approaches. Most notably, matrix-vector multiplication has demonstrated considerable efficacy in recent implementations using in-memory computation techniques. However, when shifting gears to the more complex inverse matrix-vector multiplication, additional challenges surface, notably due to the heightened complexities involved in circuit design and implementation. Understanding these hurdles is critical, as they underline the significance of the latest achievements reported in the field.

In an exciting development, researchers have unveiled an integrated analogue closed-loop in-memory computing accelerator explicitly designed for performing inverse matrix-vector multiplication. This cutting-edge chip leverages static random-access memory (SRAM) technology, meticulously fabricated using 90-nanometer complementary metal-oxide-semiconductor (CMOS) technology, demonstrating a remarkable fusion of innovation and practical applicability. The architecture comprises two 64 × 64 memory arrays, creating a robust mechanism for executing algebraic manipulations, vital for various computational tasks.

The heart of this chip lies in its analogue feedback loop, a sophisticated arrangement that incorporates essential components such as operational amplifiers, digital-to-analogue converters (DACs), and analogue-to-digital converters (ADCs). This intricate configuration not only enhances the performance of the chip but also enables it to process data with unmatched efficiency. The inclusion of operational amplifiers allows for precise control over the signal flow, ensuring high fidelity in processing inverse calculations, a necessity for applications requiring a high degree of accuracy.

Experiments have illuminated the versatile capabilities of this analogue computing accelerator. One particularly compelling application showcased is its utility in solving complex systems of differential equations through recursive block inversion. Such capabilities represent a leap forward in computational power, particularly in fields like control systems and dynamic modeling, where real-time data processing is critical. The chip essentially streamlines operations that would otherwise demand extensive computational resources from traditional digital systems.

Moreover, its performance extends beyond solving mathematical equations. The chip has been identified as a crucial tool for trajectory tracking in sounding rockets utilizing a Kalman filter. This algorithm, a cornerstone of modern control theory, benefits immensely from the rapid computations enabled by in-memory processing. The chip’s ability to provide rapid data analysis can significantly enhance real-time operational accuracy during challenging aerospace missions where precision is paramount.

In another enlightening application, the in-memory computing accelerator demonstrates its efficacy in accelerating inverse kinematics computations for robotic arms. In robotics, the ability to translate desired end-effector positions back into joint configurations is a challenging yet vital process. The enhanced computational capabilities of the chip offer substantial improvements in speed and efficiency, thereby enabling smoother and more responsive robotic movements. This could lead to advancements in fields ranging from manufacturing to healthcare, where robotic assistance is on the rise.

Moreover, the results obtained from this analogue system closely align with those produced by fully digital systems operating at the same integrated circuit precision. This congruence is not merely a coincidence; rather, it speaks volumes about the chip’s design and its operational effectiveness. The advantages here are multi-faceted, offering substantial reductions in latency and energy consumption. As digital technologies continue to grapple with power consumption and speed limitations, the implications of this integrated approach could redefine future computing architectures.

The research not only achieves technical milestones but also presents a promising pathway towards sustainable computing solutions. As global emphasis shifts to eco-friendly technologies, developments like this in-memory computing accelerator might pave the way for greener computing alternatives. The reduction in energy consumption without sacrificing performance surely aligns with the growing demands for sustainability in technological advancements.

Research initiatives in this arena are essential for propelling the boundaries of what is possible in computing technology. The combined capabilities of high density, speed, and energy efficiency can foster new avenues for exploring complex problem-solving that primarily rely on conventional computational paradigms. As these technologies grow in importance, they beckon researchers and industrial practitioners alike to innovate further.

In summary, the advent of a fully integrated analogue closed-loop in-memory computing accelerator reveals a significant stride toward solving one of computing’s most intricate challenges—the inverse matrix-vector multiplication. This transformative technology not only enhances computational performance but also opens up doors for applications across various domains, including aerospace and robotics. Its potential impact on future computing systems cannot be overstated, as we stand on the precipice of a new era driven by efficient and high-performance computing solutions.

The continued exploration of in-memory computing technologies provides an exciting glimpse into the future of computing. Innovations in this space are expected to accelerate at an unprecedented pace, fostering new ideas and applications that can redefine industries. As researchers delve deeper, collaborations between academia, industry, and technological innovators will play a pivotal role in advancing these capabilities further.

The understanding of integrated systems like this accelerator marks an essential milestone in computational research and development. The scientific community stands on the brink of a significant shift in how data-intensive operations are approached. With advancements like these, the challenges of tomorrow may well be addressed today through innovative in-memory computing solutions, significantly reshaping the fabric of modern technology.

Subject of Research: In-memory computing for inverse matrix-vector multiplication

Article Title: A fully integrated analogue closed-loop in-memory computing accelerator based on static random-access memory

Article References:

Mannocci, P., Zucchelli, C., Andreoli, I. et al. A fully integrated analogue closed-loop in-memory computing accelerator based on static random-access memory.
Nat Electron (2026). https://doi.org/10.1038/s41928-025-01549-1

Image Credits: AI Generated

DOI: https://doi.org/10.1038/s41928-025-01549-1

Keywords: in-memory computing, inverse matrix-vector multiplication, Kalman filter, differential equations, robotics, semiconductor technology

The study, recently published in the prestigious journal Nature Electronics, outlines a comprehensive framework for establishing carbon-neutral data centers in space. Prof. Wen Yonggang, an Associate Provost at NTU, along with a team of interdisciplinary scientists, developed this framework to tackle the growing challenges faced by terrestrial data centers, especially in densely populated urban environments such as Singapore. With land scarcity and high real estate costs, traditional data centers are increasingly being challenged to meet both capacity and sustainability goals.

The unique proposition of utilizing space for data processing centers hinges on two substantial advantages: an abundance of solar energy and the extreme cold of the vacuum of space, which can facilitate a cooling process that is unmatched on Earth. The researchers assert that by strategically positioning satellites equipped with advanced processors, it is possible to create orbital data centers that operate without generating carbon emissions, a critical factor given the projected 165 percent increase in AI-driven computing demand by the year 2030. Indeed, in Singapore, the energy consumption attributed to data centers could rise from 7 percent to 12 percent of the nation’s electricity usage within the next few years, making the development of alternative solutions essential.

Prof. Wen emphasizes that the concept of orbital data centers aligns with a vision for sustainable computing that can transform global digital infrastructure. By capitalizing on the sun’s unrestricted energy and the natural cooling provided by the chilly environment of space—around 2.7 Kelvin—these facilities could offer superior performance compared to their Earth-bound counterparts. This innovative approach strives not just for reduced operational costs but also for a significant decrease in greenhouse gas emissions, which is of paramount importance as the world grapples with climate change.

To further substantiate their claims, the research team developed a digital twin model in collaboration with a deep-tech spin-off, Red Dot Analytics, founded by Prof. Wen. This model simulated the expected power consumption, cooling efficiency, and solar energy generation potential of space-based data centers. Early findings indicated that the method by which heat would dissipate into the vacuum of space could potentially allow for more efficient operation than conventional cooling methods used on Earth, a critical consideration for high-performance computing applications.

As technological advancements make the concept more plausible, the researchers outlined two distinct models for deploying data centers in space. The first model, referred to as Orbital Edge Data Centers, involves deploying imaging or sensing satellites loaded with AI accelerators that can process data at the source, transmitting only the essential results back to Earth for further analysis. This data reduction capability not only minimizes the volume of information that needs to be transported but also significantly cuts energy consumption and latency issues faced by terrestrial systems.

The second model proposed by the NTU research team is more ambitious: Orbital Cloud Data Centers. These would involve a constellation of satellites equipped with powerful servers, high-speed broadband links, solar panels, and specialized cooling systems designed to manage complex computational tasks including scientific simulations and AI model training. This decentralized approach allows for scaling operations effectively, providing a flexible alternative to conventional data center architecture situated on Earth.

Yet, there are challenges that remain, particularly around the initial carbon footprint associated with rocket launches, which are carbon-intensive. However, the study introduces a novel metric termed life-cycle carbon usage effectiveness (CUE). This metric demonstrates that over time, the emissions generated during the launch of these orbital facilities could be effectively negated by the immense benefits of their operational capabilities. Factors like the development of reusable rockets and improved launch technologies are critical in paving the way forward, making space-based operations not only more feasible but also more sustainable.

Additionally, advancements in technology have made significant strides. Companies like AMD have pioneered the production of space-grade processors, while NTU’s spin-off, Zero Error Systems, develops fault-tolerant semiconductor technologies that ensure consumer-grade hardware can perform reliably in the harsh conditions of space. These innovations set the stage for achieving a long-term vision where orbital data centers become an integral part of global computing infrastructure, devoid of the limitations imposed by terrestrial constraints.

The innovative research from NTU symbolizes a collaborative spirit between academic institutions and the tech industry, vital for addressing the pressing challenges of our time. With the invaluable input from Prof. Louis Phee, NTU’s Vice President of Innovation and Entrepreneurship, the project highlights the importance of fostering creativity and interdisciplinary synergy among researchers and entrepreneurs alike. This research marks a pivotal step toward a future where sustainable computing solutions can be effectively developed and implemented.

NTU Singapore has leveraged its strong foundation in research and technological development to position itself as a leader in sustainability and advanced computing solutions. The study encapsulates NTU’s vision to mitigate the challenges of data consumption while simultaneously addressing environmental concerns through innovation.

In conclusion, the ambitious project proposes a viable and transformative solution to the challenges of modern computing. By breaking away from traditional methods and exploring the unbounded opportunities available in space, researchers at NTU are pioneering a future where sustainable and high-performance computing can coexist, ensuring that the growing digital needs of society can be met without compromising the integrity of our planet.

Subject of Research: Carbon-neutral data centres in space
Article Title: The development of carbon-neutral data centres in space
News Publication Date: 27-Oct-2025
Web References: 10.1038/s41928-025-01476-1
References: [1] Goldman Sachs Research, [2] Infocomm Media Development Authority, [3] Singapore Business Review, [4] AMD, [5] Zero Error Systems
Image Credits: NTU Singapore

Keywords

Space data centers, sustainable computing, carbon-neutral technology, artificial intelligence, energy efficiency, orbital systems, environmental science.

Recent developments in this field have led to the emergence of a groundbreaking 64-kilobit non-volatile digital compute-in-memory macro, specifically designed for artificial intelligence applications. Built on 40-nanometer spin-transfer torque magnetic random-access memory technology, this innovation marks a significant leap forward, addressing many of the limitations that plagued earlier generations of compute-in-memory architectures. The ability to conduct computations directly within the memory cell itself enables a drastic reduction in the amount of data transfer necessary, ultimately accelerating processing times and enhancing energy efficiency.

At the core of this revolutionary macro is its ability to perform in situ multiplication and digitization at the bitcell level. This means that rather than relying on external computing components, the macro can execute multiplication directly within the memory, thereby minimizing latency and improving speed. Furthermore, it offers precision-reconfigurable digital addition and accumulation capabilities at the macro level, allowing for flexible and adaptive computing solutions that can cater to various application scenarios. This flexibility is particularly vital in the realm of artificial intelligence, where the precision of calculations can significantly impact model performance.

One of the key advantages of this new CIM macro lies in its support for a lossless approach to matrix-vector multiplications. This is essential for many machine learning tasks wherein maintaining data integrity during operations is crucial. The macro can handle flexible input and weight precisions, offering configurations ranging from 4-bit to 16-bit precision. Such versatility enables researchers and practitioners to fine-tune their models, optimizing them for specific tasks or hardware constraints without sacrificing accuracy or performance.

The implications of this technological breakthrough extend beyond mere computational efficiency. In practical terms, it has been demonstrated that the macro can achieve software-equivalent inference accuracy for well-known neural network architectures. For instance, when applied to residual networks, the macro maintains an impressive inference accuracy at 8-bit precision, showcasing its capability to execute complex AI models without significant downgrades in performance. Similarly, for physics-informed neural networks, it attains high fidelity in processing results at 16-bit precision, underlining its robustness across various applications.

Speed is another critical aspect where this digital compute-in-memory macro excels. When evaluating its performance metrics, it boasts computation latencies ranging from 7.4 to 29.6 nanoseconds. This is an extraordinary feat, considering that rapid processing times are fundamental for real-time applications, particularly in fields such as autonomous vehicles, real-time data analysis, and complex simulations. The rapid computation capacity will likely play a vital role in the deployment of advanced AI systems across diverse sectors.

Moreover, energy efficiency is a prominent feature of this macro. With energy efficiencies measured at between 7.02 and 112.3 tera-operations per second per watt for fully parallel matrix-vector multiplications, the macro sets a new standard in the realm of computational power. This makes it not only a potent option for large-scale AI deployments but also a more sustainable choice amidst growing concerns about the energy consumption of technological infrastructures.

The development of this CIM macro is indicative of a broader trend within the tech industry, which is increasingly prioritizing hybrid systems that meld different computing paradigms. By merging the benefits of both non-volatile memory and compute-in-memory design, this architecture represents a shift towards a more integrated approach in chip design. Such integration can potentially lead to a new generation of computing devices that perform not just with speed and efficiency but also with greater intelligence.

The design methodology behind this macro includes a toggle-rate-aware training scheme at the algorithm level, a sophisticated mechanism that allows for optimization at every stage of computation. This aids in reinforcing the macro’s accuracy while simultaneously enhancing its overall functionality. By ensuring that all components of the architecture are aligned optimally, this training scheme provides a comprehensive framework for deploying robust AI solutions.

As industries worldwide continue to explore the implications of artificial intelligence, innovations such as this non-volatile compute-in-memory macro will undoubtedly shape the future of computing technology. The seamless integration of memory and processing capabilities offers a transformative pathway to unlocking higher performance levels while managing inherent limitations associated with traditional methods.

In conclusion, the advancements represented by this non-volatile compute-in-memory macro signify a major breakthrough in artificial intelligence and computing. It not only addresses the ongoing challenges of speed and energy efficiency but does so while maintaining performance integrity across various levels of precision. As this technology matures, it could pave the way for more agile AI systems that are capable of meeting the demands of future applications, ultimately leading to smarter, more responsive environments.

Technology is advancing at a breakneck speed, making it imperative for researchers and practitioners in the field of AI to stay on the cutting edge of innovation. This non-volatile CIM macro is a reminder of the exciting possibilities that lie ahead as the boundaries between memory and processing blur. By adopting such paradigms, the tech industry can not only enhance computational capabilities but also contribute to the responsible and sustainable evolution of artificial intelligence technology.

As we look forward, the importance of developing efficient, powerful, and accurately functioning AI systems cannot be understated. The emergence of this CIM macro is a testament to human ingenuity, a leap into a future where the potential of artificial intelligence can be fully realized through smart innovations in architectural design.

With continuous research and development, we may witness even more extraordinary advancements that redefine the landscape of computation. This non-volatile compute-in-memory macro stands as a potent example of where technological innovation meets practical application, offering a glimpse into the ways we will compute, learn, and interact with technology in the years to come.

Subject of Research: Non-volatile digital compute-in-memory macro for artificial intelligence applications.

Article Title: A lossless and fully parallel spintronic compute-in-memory macro for artificial intelligence chips.

Article References:

Li, H., Chai, Z., Dong, W. et al. A lossless and fully parallel spintronic compute-in-memory macro for artificial intelligence chips.
Nat Electron (2025). https://doi.org/10.1038/s41928-025-01479-y

Image Credits: AI Generated

DOI: Not provided.

Keywords: Non-volatile compute-in-memory, artificial intelligence, spin-transfer torque magnetic random-access memory, digital computing, matrix-vector multiplication, energy efficiency, computational latency.

The memristor, a two-terminal component whose resistance changes based on the history of voltage and current, has been a subject of intense research due to its promise for non-volatile memory and neuromorphic computing. By incorporating optoelectronic properties into polymer-based materials, the research team transcended traditional electrical memristance, enabling the device to dynamically respond not only to electrical stimuli but also to optical signals. This dual-modality represents a significant leap forward, particularly for edge computing devices that require swift, localized decision-making with minimal energy consumption.

A fundamental challenge addressed in this work is the power consumption and computational latency inherent in conventional sensor-to-processor architectures, where data must be transmitted to centralized units for analysis. The newly engineered polymer memristor offers real-time sensing and processing capabilities, leveraging its photoresponsive characteristics to directly convert incident light information into modulated memristive states. This integration drastically reduces the need for data movement, which is often the primary source of energy inefficiency in edge systems.

The researchers utilized an optoelectronic polymer matrix embedded with nanostructures that promote strong photo-induced charge separation and transport, essential for the memristive behavior under light exposure. This hybrid molecular design ensures that the device exhibits multi-level resistance states controllable via both electrical voltage pulses and optical inputs. Such tunability affords a versatile platform capable of implementing complex logic and memory functions, tailored dynamically during operation.

Notably, the memristor maintains a high endurance and stability across thousands of switching cycles, a critical attribute for practical deployment. The dynamic control of the device’s conductance states enables precise modulation of its electrical properties, effectively allowing the encoding and retention of information with a power envelope far lower than traditional semiconductor components. This characteristic positions the polymer memristor as a promising candidate for sustainable electronics in low-power Internet of Things (IoT) applications.

The device architecture supports in-sensor edge computing where information processing is embedded directly within the sensory units, bypassing the need for extensive off-chip computation. This architectural paradigm aligns with the growing demand for smart sensors capable of instantaneous data interpretation, facilitating faster response times in applications such as autonomous vehicles, wearable health monitors, and smart surveillance systems.

Moreover, the optical stimuli that control the memristor states open avenues for integrating optical communication channels into edge devices. This compatibility facilitates the development of hybrid systems that combine electronic and photonic functionalities, enhancing signal processing speeds and bandwidth. The inherent flexibility of the polymer-based system also suggests potential for integration with flexible electronics and conformable devices, broadening the scope of application environments.

The research team demonstrated that through precise manipulation of voltage and light intensities, the memristor could simulate synaptic functions akin to those found in biological neural networks. By emulating short-term and long-term plasticity, the device showcases its potential role in neuromorphic computing architectures that model cognitive processes with remarkable energy efficiency.

Key experimental results included the characterization of the memristor’s current-voltage behavior under varied illumination conditions, revealing distinct photo-induced resistive switching with fast response times. The multi-level resistance modulation was systematically controlled, highlighting the device’s capacity for complex data storage and retrieval within a compact footprint. Such performance metrics are critical for scalable edge computing solutions where physical space and energy budgets are constrained.

Beyond functionality, the choice of polymer materials conveys significant advantages in terms of cost-effectiveness, ease of fabrication, and environmental friendliness compared to traditional inorganic semiconductor devices. The solution-processable nature of these polymers facilitates room-temperature manufacturing, potentially enabling roll-to-roll production techniques that are indispensable for mass-market deployment.

The implications of this work extend beyond immediate applications, posing transformative prospects for the broader field of optoelectronics and smart materials. By marrying memristive behavior with optoelectronic responsiveness in a dynamic, controllable manner, the study lays a foundation for next-generation devices that could redefine computing paradigms, pushing intelligence to the very edges of sensor networks.

Future research directions anticipated from this breakthrough include optimizing the spectral response range of these polymer memristors to accommodate diverse lighting environments and exploring three-dimensional device architectures for enhanced integration densities. Additionally, refining the interplay between electrical and optical control signals may unlock unprecedented levels of computational complexity and adaptability in real-world scenarios.

In summary, the development of optoelectronic polymer memristors with dynamic control heralds a new era of power-efficient in-sensor edge computing, marrying cutting-edge materials science with innovative device engineering. This synergistic advance holds the promise to dramatically reduce the energy footprint of pervasive computing technologies while enhancing their responsiveness and intelligence, marking a significant stride toward pervasive, sustainable digital ecosystems.

Article References:
Zhou, J., Li, W., Chen, Y. et al. Optoelectronic polymer memristors with dynamic control for power-efficient in-sensor edge computing. Light Sci Appl 14, 309 (2025). https://doi.org/10.1038/s41377-025-01986-9

Image Credits: AI Generated

DOI: https://doi.org/10.1038/s41377-025-01986-9

Breaking new ground, a research team led by the Singapore University of Technology and Design (SUTD) has presented an innovative methodology that circumvents this limitation. In a landmark study published in the journal Materials Horizons, the team unveiled how an altermagnetic bilayer system could be driven by an external electric field to control spin polarization. This groundbreaking approach holds the potential to redefine the landscape of spintronic devices, equipping them with capabilities that could transform the future of computing.

At the core of this transformation lies the intriguing concept of altermagnetism, a unique form of magnetism wherein the spins of electrons within a material orient in opposite directions. This distinct arrangement results in the cancellation of any substantial macroscopic magnetization, a characteristic that sets altermagnetism apart from conventional magnetic materials like ferromagnets and antiferromagnets. The dual-spin configurations allow for non-collinear spin currents, making altermagnetic materials particularly suited for applications in advanced spintronic technologies where precise spin manipulation is paramount.

The team’s experiments focused on a bilayer system composed of ultra-thin layers of chromium sulfide (CrS), a material recognized for its altermagnetic properties. Through meticulous application of an electric field, the researchers discovered that the spin polarization could be fully reversed, achieving an impressive spin polarization rate of up to 87% at room temperature. This remarkable finding showcases the potential for all-electrical manipulation of spin states in a practical, room-temperature environment, which could lead to a new generation of compact and efficient spintronic devices.

The pivotal mechanism behind this discovery centers around what the researchers term "layer-spin locking." In their bilayer structure, each layer can conduct currents of opposite spin polarization independently. When an electric field is applied, it selectively modifies the energy levels in the structure, allowing for finely tunable spin-polarized currents. The experimental setup resembles two conveyor belts operating at different speeds, each carrying electrons with opposing spins. By varying the voltage applied, one layer can dominate over the other, effectively flipping the spin state of the electrons and providing unprecedented control over spin currents.

The excitement surrounding this finding is palpable. Dr. Rui Peng, the lead author of the study, emphasizes the significance of controlling spin solely via electrical means, a breakthrough that could eliminate the complications of integrating magnetic fields in small-scale devices. This finding brings to life a vision of ultra-compact spintronic devices that promise higher efficiency and performance, marking a significant stride towards realizing functional applications in everyday technology.

The implications of this research extend far beyond theoretical exploration. The potential applications are vast, including next-generation computing systems that rely upon rapid data processing and memory storage capabilities, as well as innovations in quantum technologies that demand precise control over quantum states. The altermagnetic materials and methods put forth by this research could stimulate novel approaches in material design and integration strategies for spintronic devices.

As exciting as this discovery is, the research team is already laying the groundwork for further investigations. The next step involves experimental validation and the prototyping of devices that harness this newly discovered ability to control spin with electric fields. The researchers are delving into the possibilities of integrating their bilayer system with real-world electronic circuits to demonstrate its feasibility within commercial applications.

The ambition behind this work is clear: to develop practical, manufacturable spintronic devices that can outperform the capabilities of existing silicon-based electronics. Assistant Professor Yee Sin Ang, who co-led the research, underscores the potential of this study to serve as a blueprint for transforming the landscape of modern computing. With a focus on efficiency and speed, the team is poised to make a significant impact in a field characterized by rapid advancement and technological promises.

The urgency to develop such transformative technologies could not be higher. As the world grapples with the demand for faster computing and increased energy efficiency, all-electrical spintronics emerges as a central player in the unfolding narrative of technological innovation. This research stands as a testament to the pioneering spirit of scientists working at the intersection of materials science and engineering, illuminating a promising path toward ultra-fast, energy-efficient computing.

Moreover, the collaborative efforts between SUTD and other esteemed institutions—including the Hong Kong University of Science and Technology, Beijing Institute of Technology, Zhejiang University, and A*STAR Singapore—highlight the global nature of this scientific endeavor. By pooling expertise from various sectors of research, the team maximizes its potential to realize breakthroughs that can yield real-world benefits.

This research not only represents a major step forward in understanding and leveraging altermagnetism for spin control but also places the spotlight on the importance of interdisciplinary collaboration in addressing pressing technological challenges. As the quest for next-generation spintronic devices unfolds, the innovations developed in this study may serve as a catalyst for future advancements in electronics that redefine how we understand and utilize information storage and processing.

In summary, the team’s pioneering research into altermagetic bilayers and the all-electrical control of spin currents signifies a watershed moment in the evolution of spintronics. The prospects for integration into practical applications underscore the potential to transform computing philosophies, paving the way for a future characterized by exceptional efficiency and unprecedented performance.

With the momentum building around these discoveries, the global scientific community eagerly anticipates the outcomes of ongoing research and prototype development in the realm of altermagnetic materials. As this field continues to evolve, it may very well shape the future of electronics as we know it.

Subject of Research: Electric field control of spin polarization in altermagnetic bilayers
Article Title: All-Electrical Spin Control: The Rise of Altermagnetic Spintronics
News Publication Date: October 2023
Web References: Materials Horizons
References: Not available
Image Credits: Credit: SUTD

Keywords

Spintronics, Altermagnetism, Electron Spin, Electric Field Control, Room Temperature Spintronics, Chromium Sulfide.