Currently, data centers are estimated to account got 1-2% of global power consumption, and this number is expected to rise to 3-4% by the end of the decade. For a large part, this increase can be attributed to the recent developments in AI and its use of GPUs. Even though computations on GPUs are significantly more power efficient than on CPUs, the scale of computation needed to train and apply AI models is staggering and doubles every nine months.

Neuromorphic Computing is seen as an energy-efficient alternative to GPU-based AI. This approach uses artificial neurons to do computations and is inspired by the structure and function of the human brain (which only requires about 0.3 kWh of energy per day; the equivalent of ~100 ChatGPT-4 queries).

There is currently significant investment in Neuromorphic Computing, both at fundamental level (such as materials and chip design) and applied level (algorithms, programming frameworks and applications). Several experiments chipsets exist (i.e., as Loihi 2 and SpiNNaker2) as well as large scale systems (i.e, SpiNNclould, Hala Point and DeepSouth). Many additional examples can be found on the Open Neuromorphic site. Within the Netherlands, initiatives like NL-ECO are exploring the benefits of this approach.

The goal of this project is to get an overview of the state-of-the-art in Neuromorphic Computing. What hardware is currently available? Is the software ecosystem mature enough? What applications can benefit from neuromorphic computing? Can we get access to suitable testbeds to experiment with this paradigm? What opportunities and downsides do we see?

Outcomes

Lorentz Workshop

Over the last several months, we have taken part in various initiatives, starting with the Lorentz workshop on Sustainable Computing with Neuromorphic and Quantum-Inspired Technologies that took place in January 2026. The workshop focused on bringing awareness to various use cases in science and industry, and working out what specific emerging technologies can potentially benefit those use cases. Examples included tiny embedded neuromorphic chips for inference on the edge to quantum-based true random number generators for efficient sampling.

NC-NL engagement

The workshop was followed by engagement with the newly formed Neuromorphic Computing NL (NC-NL) coalition, which aims to give strategic direction and momentum to the already thriving neuromorphic community in the Netherlands. NC-NL is pursuing three strategic action lines (ALs) that outline actionable steps towards the realisation of the overarching goals of the coalition:

• AL1 - Ecosystem development: Building a strong national foundation for neuromorphic computing.
• AL2 - Market-driven Application Lab: A dedicated application lab to accelerate commercial adoption of neuromorphic computing.
• AL3 - Prototyping Facility for Emerging Technologies: Prototyping infrastructure to accelerate the translation of scientific breakthroughs into next-generation hardware.

Currently, NLeSC is involved in progressing AL2 (Market-driven Application Lab).

SNNBook

NLeSC is also contributing content to the SNNBook effort - an open-source repository of practical knowledge about neuromorphic computing that is organised around three main topics:

• Foundations of SNNs: Covers the fundamental concepts from neuroscience that underpin the theory of spiking neural networks. The main goal is to explain how artificial neurons and synapses model the computational principles and information representation mechanisms implemented by biological neurons.
• Training SNNs: Covers the algorithms used for training SNNs, such as credit assignment, surrogate gradients and biologically inspired training. The goal in this section is to give the user practical knowledge and instruments to train their own SNNs.
• Deploying SNNs: Covers topics related to the deployment of trained SNNs on various hardware platforms, with the goal of highlighting the idiosyncrasies of different neuromorphic platforms and presenting various performance metrics.