The Basque Country, a vibrant region in Spain with a rich history of innovation and a strong industrial sector, has launched the SiliconBurmuin initiative to capture the momentum of the rise of neuromorphic technology in the global AI boom. The initiative leverages major EU and Spanish investments in semiconductors and reduced entry barriers with access to open EDA tools, royalty-free IP, and cost-effective chip production options.
By harnessing the region’s expertise in neuroscience, mathematics, AI, and digital circuit design (SoC and RTL), SiliconBurmuin aims to create a multi-disciplinary neuromorphic community in the Basque Country to address the technological challenges faced by Basque industries, which reflect global market needs. The initiative focuses on developing neuromorphic IP and chips tailored to these industries to unlock competitive advantages, while also fostering the growth of deep-tech startups to commercialise SiliconBurmuin technology on a global scale. The initiative is also committed to training the next generation of engineers in ASIC design, supporting the long-term growth of the Basque neuromorphic community.
The Five Pillars of SiliconBurmuin
Pillar A: Neurocomputing
Develop computing technologies to run biologically realistic mathematical theories, algorithms, and computational models to support neuroscience research, with a focus on vision-related biological structures and mechanisms across different species.
Pillar B: Neuromorphic
Transform neuroscience knowledge and neurocomputing artifacts from Pillar A, particularly those related to vision, into energy-efficient neuromorphic IP that can be implemented on FPGA and silicon (RTL designs) or run on neuromorphic processors (SNN models).
Pillar C: Semiconductor
Implement the neuromorphic RTL IP from Pillar B into silicon-optimised ASICs, promote its adoption across the industry, and drive commercialisation through the creation of deep-tech startups.
Pillar D: Industry
Integrate state-of-the-art neuromorphic technology –both commercial and SiliconBurmuin IP and ASICs from Pillars B and C– into industry-ready systems that deliver superior performance and efficiency for demanding vision-related applications.
Pillar E: Education
Enhance semiconductor design skills in the Basque Country through the SoC4Sensing University-Business Chip Chair at UPV-EHU, co-funded by the Spanish Government, IKERLAN, the Basque Microelectronics Hub (BMH), and SoC-e System-on-Chip Engineering. Leveraging selected SiliconBurmuin’s RTL IP and expertise, the Chair offers hands-on training for students.
SiliconBurmuin members
IKERLAN
IKERLAN is a member of the Basque Research and Technology Alliance (BRTA) and the largest R&D centre of Mondragon Cooperative Corporation with more than 400 researchers and engineers.
IKERLAN coordinates the SiliconBurmuin initiative and is designing BEGI, a licensable processor IP for multimodal DVS inputs (standard, foveated, light-field, and eventually hyperspectral). BEGI generates data-efficient structures compatible with mainstream AI models, such as depth maps, that encode the 3D surroundings, bridging the efficiency and low-latency advantages of neuromorphic sensing with the usability of AI frameworks and processors. It implements event-driven lightweight filtering and optical flow estimation modules that can be dynamically adjusted to balance accuracy, energy consumption, and latency in each operating condition. IKERLAN is also developing BEGI-EVK, an evaluation platform to demonstrate BEGI alongside other technologies from SiliconBurmuin and European partners.
ACHUCARRO & BCBL
The ACHUCARRO Basque Centre for Neuroscience is one of the Basque Excellence Research Centres (BERCs) dedicated to research in neuroscience with a focus on neuron-glia interactions.
The Basque Centre on Cognition, Brain and Language (BCBL) is also a BERC dedicated to studying cognition, brain function, and language.
In SiliconBurmuin, ACHUCARRO and BCBL collect and curate neuroscientific and magnetic resonance imaging datasets from human brain regions involved in relevance assignment.
This data forms the foundation for understanding the mechanisms used by the brain to allocate attentional resources to complex environmental stimuli and is being used to train SNNs with the aim of capturing the same cognitive effects on resource-limited neuromorphic chips.
BCAM
In SiliconBurmuin, BCAM is developing mathematical theories and computational models (i.e., Neural Automata) derived from neuroscientific observations and experimental datasets curated by BCBL and ACHUCARRO to enable the design of biologically plausible neuromorphic circuits and SNN training.
IIS BIOBIZKAIA
In SiliconBurmuin, IIS BioBizkaia develops computerised tests that use the high-temporal resolution of DVS sensors to capture and evaluate visual attention and high-speed micro-saccades as potential biomarkers for neurodegenerative diseases such as Alzheimer’s and Parkinson’s.
UPV/EHU
In SiliconBurmuin, UPV/EHU explores the feasibility and trade-offs of hyperspectral DVS technology building on Prophesee DVS sensors to meet the latency and robustness requirements imposed by CAF autonomous train operating in adverse weather and lighting conditions. UPV/EHU also trains SNNs using experimental datasets collected by ACHUCARRO and BCBL and designs RISC-V-based SoC infrastructure to host the SiliconBurmuin neuromorphic IP.
VICOMTECH
In SiliconBurmuin, VICOMTECH designs computer vision algorithms to process outputs from IKERLAN’s BEGI to detect objects and obstacles with minimal latency in challenging environments encountered by the CAF autonomous train, such as low-light tunnels.
TECNALIA
TECNALIA, the largest applied R&D centre in Spain and a member of BRTA, boasts a team of over 1,400 researchers. Its value proposition lies in its deep expertise in key enabling technologies, including AI.
TECNALIA designs efficient forward-only learning algorithms for event-driven neural networks, such as SNNs, and explores ways to enhance these networks with extended capabilities such as out-of-distribution detection and continual learning. The team at TECNALIA also develops post-hoc explainability methods suitable for SNNs, as well as different neuromodulation techniques for different purposes such as catastrophic forgetting avoidance.
CAF
CAF is Europe’s fourth largest and Spain’s leading manufacturer of passenger trains. Its research unit, CAF R&D, focuses on exploring high-potential innovations in AI-based transportation systems.
In SiliconBurmuin, CAF conducts the validation (assess accuracy and response time) of the developed neuromorphic technology-based perception solution, including standard, foveated, hyperspectral, and light-field DVS and vision algorithms, in simulated and real-world train environments.
BMH
In SiliconBurmuin, BMH oversees the dissemination of results and objectives while identifying technology transfer and exploitation opportunities for the resulting IP. These efforts align with the strategic goals of the semiconductor ecosystem in the region.