- As AI data centers consume an increasing number of RAM chips, researchers are looking for cheaper and scalable alternatives.
- imec research is seen by many as the answer, even as engineering challenges continue to exist.
- Two published studies using ferroelectric materials show that the capacitors are essentially capable of replicating modern DRAM functionality.
It’s no secret that AI has disrupted much of the world: many are applauding the arrival of a “smart” digital assistant, while others are using it to speed up the most mundane tasks that previously required supervision.
On the other side of the equation, some find their jobs redundant, others are forced to upskill or switch to different industries just to eke out a living in 2026.
However, there is one area that has been significantly affected and whose effect may have been underestimated until now: the PC hardware industry.
DRAM for data centers first
The reason the PC hardware industry (and to a lesser extent the smartphone and game console industries), or any industry that uses fast RAM and NAND flash for its core functionality, is affected is the overwhelming demand in data centers.
With billions of dollars in construction and procurement budgets, the average consumer is unlikely to compete for memory with a data center built by hyperscalers, and rightly so: there’s simply too much money at stake here.
However, the problem for data centers is different: despite consuming almost 70% of all memory produced in 2026, resulting in the worst memory supply crisis in the last 15 years, things are only predicted to get worse, even as demand continues to grow unabated.
As memory giant Micron tells consumers to wait until 2028 before they can expect any relief in their segment, and SSDs are heading down the same path, thanks to a similar situation when it comes to high-density NAND flash-based storage, AI data centers are also looking at a memory breakthrough that could enable cheaper, faster memory in the near future to better handle their needs.
However, European chip research lab imec appears to be working on an answer. At the 2026 IEEE/JSAP Symposium on VLSI Circuits and Technology, the event showcased two important advances that could allow new ferroelectric memory to become widespread over the next decade, as it attempts to address the AI-focused memory shortage with a solution that better serves data center consumers.
At the center of current advances is a ferroelectric capacitor that operates at low voltage, allowing for a high number of write cycles and holding a charge well, making it a potentially viable replacement for traditional DRAM-based memory.
The second is a transistor that they managed to stack vertically, allowing for a denser NAND-flash style storage design while also adding a backdoor modification to address memory erasure issues.
FeRAM is not a new concept; First conceived in 1952, it shows great promise, although most researchers agree. Until recently, it had limited interest, but a world spending billions on AI data centers, with increasingly scarce memory and storage resources, has brought it back to the forefront, even if most of its profits still keep it limited to laboratory conditions.
“This work shows how imec’s multidisciplinary expertise, from materials science to advanced 3D integration, allows us to address some of the most pressing challenges in memory technology,” said imec program director Maarten Rosmeulen. “We are exploring multiple paths to the memory solutions that will be needed to sustain the rapid growth of AI and data-intensive applications.”
It should be noted that imec does not work in a vacuum; He shares his research with hundreds of industry partners, including chip designer Nvidia, photolithography system maker ASML, and fabulous giant TSMC.
Other industrial partners include Intel, Samsung, Micron, Qualcomm, AMD and Apple, indicating that their research could eventually be integrated into future memory offerings by multiple interested entities, including other hyperscalers.
If imec’s research solves the modern data center conundrum by offering a denser, cheaper memory module using different materials, it could very well spark a new AI-centric war for high-end memory and storage.
But for now, it appears to require a considerable amount of time before it is ready for production, and imec admitted it was “addressing remaining challenges” while reminding interested parties that the entire trailer was still a proof of concept and still in the research stage.
Follow TechRadar on Google News and add us as a preferred source to receive news, reviews and opinions from our experts in your feeds.
