Japanese tech vendor TDK has unveiled a breakthrough in photo-spintronic conversion technology that would enable 10 times faster data transmission, which could support the ultra-fast data transfer required for the development of artificial intelligence (AI).
Touted by TDK as “the world’s first Spin Photo Detector,” the device amalgamates optical, magnetic and electronic elements to enable a data response time of 20 trillionths of a second. This essentially instantaneous transmission is an advance over the best semiconductor-based photo detectors that move data between graphics processing units (GPUs) in today’s AI applications.
TDK builds batteries for the iPhone, but for this photonics advance, the company reinvented its magnetic heads technology using its magnetic tunnel junction (MTJ), which is already deployed in countless hard disk drives to support photonics. (In commercial applications, photonics often refers to the science of data transfer using light. The “photo” in Spin Photo Detector refers to light, not photographs.)
In comparison to legacy chip-based photo detectors, which have lower performance at shorter wavelengths, TDK’s new photo technology transfers data at remarkably high speed despite shorter wavelengths. Also advantageous, spin photo detector operates at a wide array of wavelengths, from visible light to near-infrared light, which provides a high degree of interoperability with various infrastructures.
The technology’s capacity to detect light at ultra-high speeds will enable Spin Photo Detector to drive advances in uses such as virtual reality smart glasses and an array of applications that use high-speed sensors. Due to their powerful resistance to cosmic rays (high-energy particles that fly through space at nearly the speed of light), it’s possible that this technology will be used in space travel.
Perhaps its most significant use will be in AI. Current electronic circuitry is a major bottleneck holding back the speed at which even the fastest GPUs transfer data from chip to chip. TDK’s new advance eliminates that bottleneck by moving data at faster speeds than ever before.
Another disadvantage of electronic circuitry is that the speed of transmission decreases with distance; optical interconnects maintain their transmission rates regardless of distance. TDK’s new technology is also more energy efficient than electronic interconnections, offering a key improvement for the energy-hungry AI sector. Consequently, if data centers are to keep pace with the data transfer needs for AI’s growth, the shift to this type of optical technology is essential.
TDK’s Spin Photo Detector technology is still nascent, and the idea of constructing a series of partnerships with vendors of integrated circuits is a lengthy and expensive task. Still, TDK foresees the technology to ultimately be less expensive because of its need for fewer wafer processes. The photonic integrated circuit market is currently enjoying torrid growth, driven by the massive data transfer requirements of generative AI. Research firm IDTechEX predicts that the photonic integrated circuit market will grow ten times larger over the next decade, to $54 billion.
TDK has demonstrated that the technology works in early tests with Nihon University in Japan. It plans more tests and expects to offer the Spin Photo Detector for testing to customers by March 2026. The company hopes to launch full production within three to five years.
