Japanese electronics giant TDK has announced a major breakthrough in optical technology that it claims could eliminate one of the most significant hurdles in the development of generative artificial intelligence: high-speed data transfer between processors.
In a development reported by the Financial Times, TDK said it has created the world’s first “spin photo detector,” a device that integrates optical, electronic, and magnetic components to enable unprecedented response times of just 20 picoseconds—or 20 trillionths of a second. The innovation could offer data processing speeds ten times faster than current semiconductor-based photo detectors.
Hideaki Fukuzawa, senior manager at TDK’s next-generation products development centre, explained that the current bottleneck in AI systems lies not with GPU computing power, but with the slow transfer of data between processors. “This data transfer is the biggest bottleneck for AI rather than the semiconductor GPU performance,” he said. “Since we can break through many of the current bottlenecks, we think this technology will be a game-changer for the AI and data centre industry.”
The new detector, tested by Professor Arata Tsukamoto of Nihon University, operates by transferring data with light rather than electrical signals, significantly increasing speed while reducing power consumption—an essential consideration as AI workloads continue to escalate.
TDK, known for supplying batteries to Apple, adapted its magnetic head technology—originally developed for hard disk drives—to engineer this new photonics solution. The company aims to provide sample units to partners by March 2026, with mass production expected within three to five years.
Despite challenges in developing a compatible ecosystem with integrated circuit designers, TDK believes its spin photo detector will be more cost-effective than current solutions due to streamlined wafer processing.
The announcement comes as demand for faster, more energy-efficient AI infrastructure grows. The global market for photonic integrated circuits is projected to surge to $54.5 billion over the next decade, driven largely by generative AI, according to tech research firm IDTechEx.
TDK’s innovation enters a competitive field where major players, including Nvidia and Taiwan Semiconductor Manufacturing Company (TSMC), are aggressively pursuing similar advances. Nvidia’s $7 billion acquisition of Mellanox Technologies in 2020 underscored the industry’s urgency to resolve data transfer limitations within AI systems.
Beyond AI and data centers, TDK sees additional applications for its technology in emerging markets such as smart glasses for AR/VR and ultra-fast image sensors.
If successful, TDK’s breakthrough could place the company at the forefront of next-generation computing, reshaping how data is transferred and processed across a rapidly evolving tech landscape.
