💎 A memory chip that keeps your phone running for two weeks on a single charge. A semiconductor that works even in outer space. And its secret ingredient? Diamond. This isn't science fiction. It is the cutting edge of next-generation chip technology being developed right now by Japanese universities and companies.

Japan's "Power Revolution" for an AI-Hungry World

The explosive growth of artificial intelligence has thrust the semiconductor industry into an unprecedented crisis: energy consumption. Data centers powering generative AI training and inference are consuming staggering amounts of electricity, with some projections suggesting AI-related energy use could account for nearly 10% of global power consumption by 2030.

Two next-generation chip technologies emerging from Japan offer compelling solutions to this challenge. The first is MRAM (Magnetoresistive Random Access Memory), an ultra-low-power memory technology. The second is diamond semiconductors, dubbed the "ultimate semiconductor material." Both technologies ranked highly in Nikkei XTECH's "Technology Future Investment Index," signaling growing confidence from investors and industry leaders alike.

MRAM: The Memory That Never Forgets

How It Achieves 1/100th the Power Consumption

MRAM, short for Magnetoresistive Random Access Memory, stores data using the magnetic spin of electrons rather than electrical charges. Think of it as recording information with tiny magnets. Since the north-south orientation of a magnet doesn't change when you turn off the power, data is preserved indefinitely without consuming any energy.

This stands in stark contrast to today's mainstream memories, DRAM and SRAM, which require constant electrical power just to hold onto data. The moment you cut the power, everything is lost. MRAM eliminates this fundamental waste, reducing standby power consumption to virtually zero.

How big is the impact? PowerSpin, a startup born from Tohoku University in Sendai, has verified that integrating MRAM into an AI processor can reduce power consumption by up to 2,000 times. In practical terms, this could mean smartwatches that last weeks instead of days, and IoT sensors that run for years without battery replacement.

Japan's World-Leading MRAM Research

The global leader in MRAM research is Tohoku University's Center for Innovative Integrated Electronic Systems (CIES), headed by Professor Tetsuo Endoh. Since its founding in 2012, CIES has produced a steady run of results.

The most recent, announced on May 20, 2025, involves SOT-MRAM (Spin-Orbit Torque MRAM), an evolution of the more established STT-MRAM technology. Built on 300mm-wafer CMOS technology, the device holds its 0.35-nanosecond write speed and 10-year data retention while cutting write power by 35% against previous designs. Write energy came in at 156 femtojoules per operation, the lowest reported for this class of memory element.

On the industry side, Tohoku University startup PowerSpin is leading commercialization efforts, modeling itself after the UK's ARM by providing MRAM-related circuit design intellectual property to chipmakers worldwide. In the automotive sector, PowerSpin has partnered with Aisin to develop fast-booting vehicle electronics. On the manufacturing side, it signed a deal with Taiwanese foundry PSMC in February 2024 targeting MRAM volume production by 2029, most likely on the second-phase line of the plant PSMC is building with SBI Holdings in Ohira, Miyagi Prefecture.

Major Japanese companies including Kioxia and Renesas Electronics are also pursuing MRAM as a next-generation memory technology. Professor Endoh expects MRAM to start displacing some DRAM between 2028 and 2030.

A Booming Market and Intensifying Competition

According to India-based Fortune Business Insights, the global MRAM market was valued at approximately $3.2 billion in 2024. By 2032, it is projected to reach $17 billion, growing at over 30% annually.

On the manufacturing front, Taiwan's TSMC, South Korea's Samsung Electronics, and America's GlobalFoundries have all established 22nm MRAM production processes. A particularly significant development came from China: Shanghai Siproin Microelectronics began shipping China's first commercial STT-MRAM chips, manufactured by SMIC (the country's largest foundry). TechInsights analysts noted that this positions China as an emerging competitor in the global MRAM landscape.

Japan leads in fundamental research and circuit design, but overseas players hold the advantage in mass manufacturing capability. Bridging this gap between "research strength" and "industrial strength" is Japan's most pressing challenge.

Diamond Semiconductors: When Gemstones Become the Ultimate Chip Material

Why Diamond Earns the Title "Ultimate"

Silicon has reigned as the king of semiconductor materials for decades. More recently, silicon carbide (SiC) and gallium nitride (GaN) have emerged as next-generation alternatives. But diamond, at least in theory, surpasses them all.

Diamond is called the "ultimate semiconductor material" for four key reasons. First, its bandgap, the energy barrier that controls whether electricity flows, is about five times larger than silicon's, enabling operation at extreme voltages. Second, its dielectric breakdown field is roughly 30 times greater than silicon's, meaning it can withstand far higher voltages without failing. Third, its thermal conductivity is the highest of any solid material, efficiently dissipating heat even under heavy loads. Fourth, it exhibits exceptional radiation resistance, making it stable in harsh environments like outer space.

In plain language: diamond can handle extreme heat, extreme voltage, extreme radiation, and still keep working. It's as close to an ideal semiconductor material as physics allows.

Japanese Institutions at the Forefront

Japan has led diamond semiconductor research since the 1980s, with institutions like NIMS (National Institute for Materials Science) and AIST (National Institute of Advanced Industrial Science and Technology) establishing the foundational technologies for crystal synthesis and device fabrication.

The most closely watched current work comes from Professor Makoto Kasu at Saga University. In 2023, his team built the world's first electronic circuit incorporating a diamond semiconductor. On December 8, 2025, working with JAXA (Japan Aerospace Exploration Agency) and others, the group reported a power-gain cutoff frequency of 120 GHz alongside an off-state breakdown voltage of 4,266 V, enabled by 157-nanometer T-shaped gate electrodes patterned with electron-beam lithography. That performance spans the microwave (3–30 GHz) and millimeter-wave (30–300 GHz) bands, putting 6G and satellite communications within reach.

Professor Kasu founded Diamond Semiconductor Inc. in February 2025, and the company began manufacturing and selling sample products in January 2026, the world's first commercial shipment of diamond semiconductor devices.

On the substrate manufacturing side, Orbray (formerly Adamant Namiki Precision Jewel) has established mass production of 2-inch diamond wafers and is working toward 4- and 6-inch sizes. Meanwhile Okuma Diamond Device, a Hokkaido University spinout based in Sapporo, has raised roughly ¥6.7 billion (about $45 million) including subsidies, and held the completion ceremony for its Fukushima plant in Okuma, Fukushima Prefecture, on May 29, 2026. The company calls it the world's first factory with a line dedicated to diamond semiconductor R&D and manufacturing, with capacity for several hundred thousand units a year. Targeting full operation by fiscal 2028, it aims to commercialize radiation-resistant devices for the decommissioning of the Fukushima Daiichi nuclear power plant.

Remaining Challenges

The main hurdles for diamond semiconductors are manufacturing cost and substrate size. Diamond is extraordinarily hard, making precision machining difficult, and current wafer sizes remain much smaller than the 300mm (12-inch) silicon wafers that are industry standard. Costs are also significantly higher than silicon.

The gaps are closing, though. Substrate availability for researchers has improved considerably, and several startups are racing to establish mass production. The Electronic Device Industry News reports that prototype devices aimed at practical use could emerge within a few years, with commercialization in selected applications plausible around 2030.

The Race with the US and China: Can Japan Bridge Research to Industry?

In MRAM, overseas foundries like TSMC, Samsung, and GlobalFoundries already have established manufacturing lines, and China has entered the field through SMIC. Japan's fundamental research at Tohoku University is world-class, but building the ecosystem to translate discoveries into volume manufacturing is an urgent priority.

In diamond semiconductors, Japan's lead is clearer. Saga University's Professor Kasu, AIST, and Orbray are among the global front-runners, with multiple commercialization initiatives already underway. Still, foreign interest is growing. Taiwanese and Chinese companies have reportedly approached Japanese researchers about collaboration, raising concerns about technology leakage amid geopolitical tensions.

Japan's strengths lie in decades of accumulated basic research and a supply chain that includes world-class materials and equipment makers. Both MRAM and diamond semiconductors are technologies that Japan pioneered and nurtured. But as one AIST researcher notes, "universities and research institutes alone cannot achieve full-scale societal implementation." Stronger government support and more active private-sector involvement will be key to converting Japan's technological advantage into industrial competitiveness.

Japan is accelerating the commercialization of next-generation chip technologies through university-born startups and national research programs. As MRAM and diamond semiconductors take on global challenges in AI power consumption, 6G communications, and space exploration, what is the state of next-generation semiconductor research in your country? We'd love to hear your thoughts.

References