SPICE · Project

Ultra-Fast, Ultra-Low-Energy Memory Chips Using Light Instead of Electricity

digitalPrototypeTRL 4Thin data (2/5)

Imagine your computer's memory works like a library where every book must be carried by hand — slow and tiring. SPICE figured out how to use tiny flashes of light to write data onto magnetic memory, like replacing those librarians with laser beams. The result is memory that writes 1,000 times faster while using 100 times less energy. They actually built a small working chip that proves this idea works, combining light-based circuits with magnetic storage on a single platform.

By the numbers

1,000x

faster write speed than state-of-the-art spintronic memory

100x

lower energy consumption than current spintronic memory technologies

600 fJ

energy per bit write efficiency

8-bit

photonic-spintronic memory demonstrator built

consortium partners across 4 countries

The business problem

What needed solving

Data centers and high-performance computing systems face a growing crisis: processors are getting faster, but memory cannot keep up — creating a bottleneck that wastes energy and limits performance. Current memory technologies force a painful tradeoff between speed, energy consumption, and cost. The industry needs a single memory technology that can serve all applications from fast cache to long-term storage without burning through power budgets.

The solution

What was built

The team built three key hardware demonstrators: a silicon photonic chip with a distribution network for switching sub-picosecond optical pulses, a silicon photonic chip with coupling structures to connect to spintronic chips, and an 8-bit photonic-spintronic memory demonstrator combining both technologies via hybrid integration. Together, these prove that light can be used to write magnetic memory at 600 fJ per bit.

Audience

Who needs this

Hyperscale data center operators (Google, Amazon, Microsoft) seeking energy-efficient memoryMemory chip manufacturers (Samsung, SK Hynix, Micron) exploring next-generation technologiesHigh-performance computing centers running exascale simulationsTelecom equipment makers needing ultra-fast RF componentsIndustrial sensor companies looking for faster, lower-power sensing platforms

Business applications

Who can put this to work

Data Center Infrastructure

enterprise

Target: hyperscale data center operators and cloud service providers

If you are a data center operator struggling with rising energy bills and processor-memory bottlenecks — this project developed a photonic-spintronic memory platform that writes data 1,000 times faster at just 600 femtojoules per bit. This could enable petabit-per-second processor-memory bandwidths your future workloads will demand. The technology targets energy-efficient exascale computing with a significantly reduced carbon footprint.

Semiconductor Manufacturing

enterprise

Target: memory chip manufacturers and foundries

If you are a semiconductor company looking for the next generation of memory technology beyond current MRAM — this project built and tested a spintronic-photonic integration platform designed to be compatible with existing electronics fabrication processes. They demonstrated an 8-bit optically switched memory at 600 fJ per bit write efficiency. This opens a path toward Universal Memory that serves all applications from cache to storage.

Telecommunications & Sensing

mid-size

Target: RF component makers and sensor manufacturers

If you are a telecom or sensor company searching for faster, more efficient components — the SPICE platform is not limited to memory. It also enables RF nano-oscillators and sensor technology built on the same spintronic-photonic integration. The consortium included industrial partners and delivered silicon photonic chips with sub-picosecond optical pulse switching, which could feed into next-generation sensing products.

Frequently asked

Quick answers

What would this technology cost to integrate into our products?

The project does not disclose unit cost data. However, the SPICE platform was explicitly designed to be compatible with future mature electronics fabrication processes, which suggests the team aimed for cost-effective manufacturing at scale. Any licensing or integration costs would need to be negotiated directly with the consortium partners.

Can this scale beyond an 8-bit demonstrator to industrial volumes?

The current demonstrator is an 8-bit proof of concept — a critical first step but far from production scale. The objective states this is designed for real-world applications beyond 2025, targeting petabit-per-second processor-memory bandwidths. Significant engineering work remains to scale from lab demonstrator to commercial product.

Who owns the IP and how can we license it?

The project was funded under Horizon 2020 FET Open (RIA), meaning IP typically stays with the consortium partners who generated it. Aarhus University (Denmark) coordinated the project. Licensing discussions would need to go through the relevant consortium members, particularly the two industrial partners.

How does this compare to existing memory technologies like MRAM?

Based on the project data, SPICE achieves 3 orders of magnitude higher write speed and 2 orders of magnitude lower energy consumption compared to state-of-the-art spintronic memory technologies. The write efficiency target is 600 fJ per bit. This positions it as a potential successor rather than a competitor to current MRAM.

When could this technology reach the market?

The project itself stated real-world applications beyond 2025. Given this was a FET Open research project that ended in 2020 with an 8-bit demonstrator, commercial deployment likely requires several more years of development, scaling, and reliability testing. Based on available project data, no commercial timeline has been confirmed.

Is there regulatory approval needed for this type of memory chip?

Memory chip technology does not face the same regulatory hurdles as biotech or pharma. Standard semiconductor industry certifications and reliability testing would apply. The team designed the platform to be compatible with existing fabrication processes, which should ease future certification paths.

Consortium

Who built it

The SPICE consortium brings together 6 partners from 4 European countries (Belgium, Denmark, France, Netherlands), with a balanced mix of 2 universities, 2 research organizations, and 2 industry partners including 1 SME. The 33% industry ratio and presence of both academic and commercial players suggests genuine interest in moving this beyond pure research. Coordinated by Aarhus University in Denmark, the consortium covers the full chain from fundamental physics through photonic chip fabrication to system integration. For a business looking to engage, the industrial partners are the most likely entry points for licensing or collaboration discussions.

AARHUS UNIVERSITETCoordinator · DK
STICHTING RADBOUD UNIVERSITEITparticipant · NL
COMMISSARIAT A L ENERGIE ATOMIQUE ET AUX ENERGIES ALTERNATIVESparticipant · FR
INTERUNIVERSITAIR MICRO-ELECTRONICA CENTRUMparticipant · BE
SYNOPSYS DENMARK APSparticipant · DK
QUANTUMWISE A/Sparticipant · DK

How to reach the team

Aarhus University, Denmark — reach out to the photonics or physics department leads involved in the SPICE project

Order a report on this consortium See AARHUS UNIVERSITET profile →

Next steps

Talk to the team behind this work.

Want an introduction to the SPICE team to explore licensing or collaboration? SciTransfer can arrange a direct connection with the right consortium partner for your needs.

Order a report on this topic More in Digital & ICT