Fraunhofer, TSRI partner on ferroelectric transistors for low-power memory for Artificial Intelligence chips

A German-Taiwanese team is developing hafnium oxide ferroelectric field-effect transistors for memory nodes smaller than 3 nm to enable computing directly in memory and cut energy use in Artificial Intelligence chips and edge devices.

Fraunhofer IPMS, Fraunhofer IMWS, and the Taiwanese research institute TSRI have launched a joint research program to develop new memory for leading chip technologies smaller than 3 nm. The project focuses on nanosheet devices built from ferroelectric field-effect transistors, FeMFETs, using hafnium oxide. These devices are presented as particularly efficient and are designed to enable computing operations directly in memory, a capability the partners say will drastically reduce energy consumption compared with conventional architectures that separate memory and compute.

The collaboration is motivated by rapidly growing demand for Artificial Intelligence and neuromorphic computing. The article identifies a key bottleneck as the transfer of data between main memory and the computing unit, which drives up latency and energy consumption in data centers and edge systems. By enabling compute-in-memory, the nanosheet FeMFET approach aims to lower both latency and energy use for workloads that are memory bound.

The research partners frame the work as laying the foundation for the next generation of energy-efficient Artificial Intelligence chips across a range of devices. Target applications cited include smartphones, automobiles, and medical devices. The project combines expertise from the two Fraunhofer institutes and TSRI to tailor ferroelectric transistor technology to sub-3 nm process nodes, with the goal of integrating low-power memory solutions into future chips for both data center and edge deployments.

62

Impact Score

Micron to exit Crucial consumer business, ending retail SSD and DRAM sales

Micron will wind down its Crucial consumer business and stop retail sales of Crucial-branded SSDs and memory after fiscal Q2 2026 (ending February 2026). The company said the move reallocates capacity to meet surging Artificial Intelligence-related demand in the data center and to prioritize enterprise and hyperscale customers.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.