Creating memory that transcends memory — paving the way for the next generation of computers.

When transferring data from one computer to another, or sharing files with someone nearby, many people use a USB flash drive. Others may have thousands of photos stored on an SD card inserted into a digital camera. Both USB drives and SD cards have become so commonplace that they can even be purchased at convenience stores. These devices belong to a category known as non-volatile memory, which retains data even when the power is turned off — a key technology supporting the future of information processing.
At the forefront of exploring the potential of non-volatile memory and advancing cutting-edge research in this field is Professor Yuichiro Mitani, from the Department of Electrical, Electronic and Communication Engineering and the Nanoelectronics Research Center, Comprehensive Research Organization at Tokyo City University.

Memory is a device that allows computers, smartphones, and other electronic equipment to record and store information temporarily or permanently. Broadly speaking, there are two types: volatile memory, which loses its data when the power is turned off, and non-volatile memory, which retains information even without power.
Professor Yuichiro Mitani conducts extensive research on this non-volatile memory. One of his major areas of focus is reliability—how to ensure data can be stored safely over long periods of time. Semiconductor devices, such as silicon transistors, are constantly exposed to electrical stress. Moreover, future technologies like quantum computers require extreme environments—operating at temperatures near absolute zero. Non-volatile memories, in particular, must retain data for decades, which means every component must withstand aging and maintain performance.
“My research asks why components degrade and how we can prevent that degradation,” Professor Mitani explains. “This is a field I’ve been pursuing since my time working in private industry.”

The amount of data generated worldwide continues to grow exponentially, with some estimates predicting it will exceed 2,000 zettabytes by 2035. Securing physical space for data storage and achieving energy efficiency have become global challenges. Moreover, with the rapid spread of artificial intelligence and generative AI, data is no longer something to simply store — it must be continually utilized and processed.
However, conventional flash memory faces structural limitations in further miniaturization, storage capacity, and processing speed. In response, a new technology called resistive random-access memory (ReRAM) has attracted increasing attention. All digital information—text, images, programs—is ultimately stored in binary form, “0” and “1.” In flash memory, these states are recorded by trapping or releasing electrons within the memory cell. In contrast, ReRAM devices use a sandwich-like structure in which an insulator is placed between two metal electrodes; “0” and “1” are recorded by changing the electrical resistance of the insulator through applied voltage. Because of its simple structure, ReRAM can be easily miniaturized and stacked, while offering high speed and low power consumption.
Despite global research efforts, many of the materials that exhibit resistive switching contain toxic, environmentally harmful, or rare substances, posing sustainability and safety concerns. Professor Mitani’s research focuses on developing next-generation ReRAM using nano-carbon materials — carbon structures engineered at the atomic level. Carbon exists abundantly on Earth, is non-toxic, and has minimal environmental impact.
“We use nano-carbon materials such as sumanene molecules — a partial structure of fullerene with a bowl shape — and graphene sheets,” explains Professor Mitani. “By sandwiching sumanene molecules between layers of graphene, we have already demonstrated resistance change when voltage is applied. Our next step is to understand the underlying mechanism and improve the device’s performance.”

Conventional computers operate by continuously exchanging data between memory and the CPU — reading data from memory, performing calculations, and writing the results back. In the age of AI and big data, however, this back-and-forth has become a major bottleneck in terms of both speed and power consumption. One promising solution is in-memory computing. Unlike conventional systems that only store data in memory, this approach allows computation to occur directly within the memory itself. By eliminating the need to transfer data repeatedly, it can dramatically reduce both energy use and processing time.
If this system becomes practical, AI hardware could emerge — devices capable of real-time, personalized, and secure processing, with highly sensitive personal data handled locally rather than on cloud servers. This concept, where information is processed and stored simultaneously, closely resembles the structure of the human brain. It has given rise to a new design philosophy known as neuromorphic computing.
Professor Mitani is working toward making neuromorphic computing a reality. The human brain learns and makes decisions through electrical signals exchanged among vast networks of neurons and synapses. His goal is to replicate this mechanism using electronic circuits. By changing the connections and strengths — the so-called weights — between artificial neurons and synapses, such systems can learn autonomously and adapt to their environment.
“This technology could enable AI that learns through sensory experience, or autonomous robots capable of independent decision-making,” Professor Mitani explains. “Today’s generative AI mimics brain-like functions through software running on traditional computers. My goal, however, is to reproduce the brain itself through hardware, using resistive switching devices as artificial neurons. There is still much to explore, and many challenges remain. But with collaboration across disciplines — and eventually with industry partners — I hope to bring these technologies into real-world application.”

After graduating from university, Professor Mitani joined Toshiba Corporation, where he was first assigned to a department researching transistors. He later transferred to a division focused on memory reliability.
“I’m the kind of person who prefers to take action rather than think too long,” he recalls. “At the company, I would use the experimental facilities to fabricate a device one day and have measurement results ready the next morning. My supervisor at the time appreciated that hands-on attitude, and that’s how I entered the world of reliability research.”
He later transitioned to an academic career for a simple reason: he wanted to keep building things himself. “In a company, as you move up, management responsibilities increase. I realized that wasn’t the kind of work that suited me — I wanted to stay close to the actual research and creation process.”

While actively conducting experiments himself, Professor Mitani has a message he hopes to pass on to his students and the younger generation.
“After experiments, I often see students discarding data from failed trials — and that’s such a waste,” he says. “All data represent truth in some form. I want them to look for what might be hidden within those results. In most cases, experiments don’t produce exactly what you expect. When that happens, I simply change my perspective and try again right away. If you act quickly and produce results fast, the path forward will always open.”

We capture scenes that move us with our smartphones, taking photo after photo. We save the words that once changed our lives on our computers, revisiting them again and again. All of these become digital data — sequences of zeros and ones — stored in invisible, microscopic rooms.
Professor Mitani’s research is an effort to make those unseen rooms more reliable, more enduring, and ultimately more human. By controlling the flow of electrons and understanding the properties of materials, his work seeks to preserve data accurately over time and make it deeply personal. In the near future, data may no longer be just information — it may become a mirror reflecting a part of the human heart.