Weebit Nano Launches In-Memory Compute Research Program

The ReRAM-based hardware development kit lets researchers test AI algorithms that compute directly inside memory.

Updated on Nov. 11, 2026 in Semiconductors

Close-up of a silicon wafer showing a regular grid of memory cells on a dark grey surface under cool, even light.
Weebit Nano Limited launched its Weebit Matrix In-Memory Compute Research Program, offering a ReRAM-based hardware kit for testing AI algorithms that compute inside memory. AI Illustration. Upload story photo >

Weebit Nano Limited has launched the Weebit Matrix In-Memory Compute Research Program, providing a ReRAM-based hardware development kit for exploring in-memory compute algorithms and systems. The company is also showing its ReRAM-based platform at electronica 2026 in Munich.

Why it matters

Moving data between processors and memory creates major performance bottlenecks and consumes substantial energy. In-memory computing, which performs calculations directly within memory, aims to reduce those data transfers.

The company projects its in-memory compute approach using Weebit ReRAM could use roughly 80 times less power than a GPU-based implementation and cut latency by almost 100 times. These figures are company projections, and the basis for them has not been independently verified.

The players

Weebit Nano Limited

Weebit Nano Limited is a developer and licensor of advanced memory technologies, including ReRAM, that is pursuing in-memory and neuromorphic computing applications.

CEA-Leti

CEA-Leti is a French research institute that has collaborated with Weebit Nano since 2019 on in-memory and neuromorphic computing research.

The details

Participating researchers will use the program's ReRAM hardware to develop and test AI algorithms, and will share ReRAM-related findings and feedback with Weebit. Weebit notes that a U.S.-based company has already used its ReRAM to evaluate a low-power AI accelerator for harsh environments, and that the company has collaborated with CEA-Leti on in-memory and neuromorphic computing research since 2019.

Timeline

  1. The IMC Research Program was launched in the week before electronica 2026.

  2. electronica 2026 runs in Munich through November 13, 2026.

The Tech Race

The launch positions ReRAM as a candidate for in-memory computing, an approach aimed at easing the processor-to-memory data bottleneck that conventional GPU architectures face. It places Weebit's technology in a competitive field of companies pursuing alternatives to power-hungry GPU-based AI hardware.

Developers and researchers can access a ReRAM-based hardware development kit to test in-memory compute algorithms. For everyday users, the practical effect depends on whether in-memory designs reach commercial products, which the announcement does not address.

The takeaway

In-memory computing is an active area of research, and ReRAM is one of several memory technologies being tested for AI workloads. Readers following AI hardware should watch whether independent benchmarks confirm the company's projected power and latency gains.

Further reading

For more on the chips behind this approach, explore Semiconductors coverage.

Source note: This article includes information reported by Electronic Design.