Skip to main content
Research

Tampere University, SemiQon and Finland–Taiwan consortium secure €3.1M SHIFT project to upscale low-power memories for data-intensive computing

Published on 28.10.2026
Tampere University
CMOS and beyond: Devices and Systems Research
Photo: Karu Films
To tackle the data center power crisis and the surging energy demands of artificial intelligence, high-performance computing and quantum systems, an international research and industry consortium has been awarded the €3.1 million "SHIFT" project (Scalable Hafnium-based Integrated Ferroelectric Technology for Data-Intensive Computing). Answering the Eureka Globalstars call, the initiative aims to scale up ferroelectric random-access memory technology to develop next-generation hardware capable of slashing global energy consumption for data-intensive computing and edge devices. 

 

The Finnish consortium is an ideal balance of academia, small and medium-sized enterprise, and a market leader in the semiconductor industry, creating a fast development timeline from excellent ideas to integrated solutions and final market-ready products. Taiwanese partners additionally bring their high-volume fabrication and packaging expertise, critical for overcoming scaling and manufacturing challenges within a short period of time.

Sayani Majumdar, Associate Professor at Tampere University

The problem: The "memory wall" and data center power crisis

Modern computing relies heavily on embedded Flash (eFlash) memory to store essential data directly on microchips. However, conventional eFlash faces severe physical scaling limits at advanced manufacturing sizes of 28 nanometers and below. Promising alternatives, such as magnetic RAM (MRAM), phase-change memory (PCM), and oxide RAM (OxRAM), require driving currents greater than 100 microamperes per memory element. This high current generates excessive heat, causes reliability issues, and requires additional cooling power that complicates chip design and manufacturing.

"AI and IOT devices operate on huge amount of data shuttling, either between processor and memory or between sensor nodes to processing units. This is the main source of huge energy consumption," Sayani Majumdar, Associate Professor at Tampere University, outlines.

"The solution that the scientists are working upon is to store and compute data on the same physical component to minimize data communication. Ferroelectric random-access memories (FeRAM) is an excellent solution that can not only store data without power but allow in-memory computing, bringing a huge energy advantage to AI and IOT."

The solution: sustainable, lead-free FeRAM

The SHIFT project addresses these scaling and energy hurdles through ferroelectric random-access memory (FeRAM). As a non-volatile memory solution that operates at lower voltage and retains stored information even when powered off, FeRAM offers significant advantages over conventional options like Flash and EEPROM: it delivers write speeds in tens of nanoseconds, read-write endurance across trillions of switching cycles, and dramatically reduced power consumption by operating with ultra-low currents through its storage elements. Beyond operational efficiency, the material composition additionally solves a critical environmental roadblock in microelectronics.

"Electronics at end-of-life generally end up in landfills, where toxic elements mix into soil, water, and air," Majumdar observes.

"Truly sustainable electronics must eliminate harmful chemicals without compromising performance or cost. Ferroelectric hafnia systems fulfill this requirement while already being compatible with standard commercial foundry lines for fast upscaling."

Key technical objectives

To demonstrate market readiness, the SHIFT consortium is pursuing broad technical goals through two main engineering milestones:

  • 256kb FeRAM Prototype: Fabricate and validate a 256-kilobit FeRAM array prototype chip based on HfZrOx processes, complete with commercial packaging and testing.

  • Cryogenic Memory Integration: Demonstrate ultra-low-power, high-performance cryogenic memories by integrating a cryo-optimized FeRAM stack directly with cryo-CMOS logic circuits.

Quantum and cryo-electronics need low-power cryo-optimized memory technology that reduces the communication needed between room temperature and low temperature, similarly to how edge computing reduces data transfer between end devices and centralized cloud infrastructure. This enables lower latency, improved energy efficiency, and more scalable quantum computing systems.

Mika Prunnila, Chief Research Officer at SemiQon

Strategic Finland–Taiwan partnership

Bringing next-generation memory hardware to market requires bridging fundamental scientific research with high-volume manufacturing. The SHIFT consortium combines specialized expertise across the global semiconductor value chain:

  • Finland: Tampere University (TAU), Applied Materials, and SemiQon contribute expertise in atomic layer deposition (ALD), thin-film and nanodevice engineering, theoretical physics modeling, and cryogenic hardware.

  • Taiwan: The Industrial Technology Research Institute (ITRI) and Lyontek Inc., working with large commercial semiconductor manufacturing companies, bring the knowledge and expertise of lab-to-fab upscaling, device testing, and packaging capabilities.

By uniting this complementary expertise for the first time, the consortium aims to accelerate the development of next-generation computing technologies.

Broad environmental and industry impact

By drastically cutting operating energy, SHIFT provides a clear pathway to reduce the carbon footprint of data centers and edge devices worldwide. At the same time, the technology opens up new possibilities for specialized computing environments, from quantum systems to deep space.

In today's data-centric world, Memory plays a critical role that goes far beyond only storing data. It has become a critical enabler for every system, from AI to connected, smart infrastructure to secure digital services and beyond. With cryogenic memories, we can bring data processing and control next to the Qubits and upscale quantum computing by eliminating data communication to room temperature processors. Similarly, autonomous systems in space will be hugely benefited by local data processing in terms of communication speed, capability and energy saving for independent decision making.

Sayani Majumdar

About SemiQon

SemiQon’s mission is to realize the promise of quantum computing by delivering scalability through powerful, resilient, and cost-effective silicon-based quantum hardware. Its technology builds upon decades of development and know-how from the semiconductor industry, making its cryo-optimized CMOS electronics commercially competitive and well-suited for mass-manufacturing. SemiQon’s first-in-the-world cryo-optimized CMOS technology delivers scalability and power efficiency to wide-ranging applications in quantum, space, and beyond. www.semiqon.com

Project funding

The SHIFT project is funded by Business Finland for the Finnish partners and the Department of Industrial Technology (DoIT) for the Taiwanese partners.

CMOS and Beyond: Devices and Systems Research Group

Author: Sujatro Majumdar