Intrinsic Semiconductor Technologies Raises Seed Funding to Develop Next-Generation Silicon Oxide Memory for Edge AI and IoT
February 21, 2025
Intrinsic Semiconductor Technologies, a UCL spinout co-founded by Professor Tony Kenyon, Dr. Adnan Mehonic, and Dr. Wing Ng, has raised £1.35 million in seed funding led by the UCL Technology Fund and IP Group to prototype its silicon oxide resistive random-access memory (RRAM) technology. The round also includes participation from UCL Business (UCLB). The capital will fund a partnership with imec, one of the world’s foremost semiconductor research and development foundries, based in Leuven, Belgium, to transfer Intrinsic’s technology to a standard CMOS process on 300mm silicon wafers — the same wafer format used in high-volume chip manufacturing worldwide.
Intrinsic was founded in 2017 to commercialise a decade of research by the team at UCL’s Department of Electronic and Electrical Engineering into memristive devices based on silicon oxide (SiOx). A memristor, also known as ReRAM or RRAM, is a two-terminal device that can switch between high and low resistance states, allowing it to store binary data. Intrinsic’s innovation lies in the material choice: by basing its memristors on silicon oxide — a material already ubiquitous in standard CMOS semiconductor processing — the company has created a memory technology that can be integrated directly onto the same chip as a processor, without requiring any new materials, equipment, or specialist fabrication processes. This is a fundamental advantage over existing embedded Flash memory, which is incompatible with modern CMOS nodes and must be manufactured separately, driving up cost, power consumption, and device complexity.
The problem Intrinsic is solving is one that is becoming increasingly acute as AI capabilities migrate from the cloud to the device edge. Microcontrollers and edge AI chips — the processing units embedded in billions of IoT devices, smart sensors, and autonomous systems — require local non-volatile memory to store model weights, configuration data, and operational parameters without power. Flash memory, the incumbent technology, is reaching the limits of what it can deliver: it cannot be scaled below certain node sizes without losing reliability, reads and writes more slowly than modern processors require, and consumes more energy than battery-powered edge devices can afford. Intrinsic’s SiOx memristors read data 10x to 100x faster than Flash, write 1000x faster, and are manufacturable using processes that already exist in every advanced chip foundry in the world.
The partnership with imec is a significant early validation. imec is routinely engaged by the world’s largest semiconductor companies — including Intel, TSMC, Samsung, and ASML — to derisk novel process technologies before industrial adoption, and its involvement signals that Intrinsic’s approach is technically credible and manufacturable at scale. The seed funding will allow Intrinsic to demonstrate switching behaviour and electrical performance at 50nm node sizes on standard 300mm wafers, generating the data required by prospective licensees in the chip design and manufacturing ecosystem.
Intrinsic’s business model is IP licensing: the company intends to establish its RRAM designs as proven, licensable intellectual property that chip manufacturers can integrate into their own products, targeting the microcontroller, edge AI, and IoT markets that together represent a £60 billion+ non-volatile memory opportunity. Subsequent to this seed round, Intrinsic raised £7 million in 2023 in a round led by Octopus Ventures, with continued backing from IP Group, the UCL Technology Fund, and £1 million from Innovate UK, to expand its engineering team and accelerate its commercialisation roadmap.
Sources
Thank you to our Patrons & Partners:
