• Innoscience supports 800 VDC rack power for AI datacenters with the industry’s most complete GaN product line from 15 volt to 1200 volt, with the highest reliability, performance, and cost-effectiveness
• GaN (Gallium Nitride) power semiconductors are critical to increase rack computing throughput and limit AI factory power consumption, due to GaN’s ability to increase both power density and efficiency

Suzhou, China, March 19th, 2026 – Innoscience (HKEX:02577.HK), the world leader in 8-inch GaN-on-Si high-performance low-cost manufacturing, announced that it received a Supplier Award from NVIDIA for its role in providing GaN for 800 VDC rack power solutions. Innoscience is proud to support 800 VDC architecture, the innovative path for the AI GPU roadmap to deliver megawatts at the rack level while increasing power conversion efficiency.

NVIDIA Senior VP Andrew Bell and Vice President Gabriele Gorla presenting Supplier Award to Innoscience

As NVIDIA’s revolutionary AI GPU roadmap increases AI computing power 1000x to 10,000x and beyond, 800 VDC rack power architecture efficiently powers megawatt-scale computing. Compared to traditional 48 volt rack power systems, 800 VDC architecture offers significant advantages in system efficiency and heat dissipation, enabling it to support AI datacenters of the future while limiting grid power consumption and minimizing CO₂ emissions.

As the world's leading GaN supplier, Innoscience's third-generation GaN devices offer exceptional high frequency, high efficiency, and high power density, providing 800 VDC architecture with a full-link GaN power supply solution, from 800 volt input to GPU terminals. With the integration of 800 VDC power architecture and GaN technology, AI datacenters can achieve the quantum leap from kilowatts to megawatts for an era of greener AI computing.

Similar to the EV industry’s move from 400 volt to 800 volt power, increasing rack voltage from 48 volt to 800 volt reduces electric current by 16x, dramatically lowering I²R power loss and minimizing copper bus bars. 800 VDC architecture requires ultra-high power density and ultra-high efficiency from 800 volt to GPU, and GaN devices are capable of simultaneously achieving these demanding requirements ~ legacy Silicon power devices cannot support AI.

Why Innoscience is the world’s leading GaN supplier:

• World-class GaN Process engineering and device design
• GaN Epitaxy developed for best performance and reliability
• World’s largest 8” Fab, over 2 billion GaN devices in production
• World’s largest and most respected power semiconductor Partners for best-in-class Supply Chain ~ wafers, components, and modules

About Innoscience

Innoscience (HKEX:02577.HK) is the global leader in GaN process innovation and power device manufacturing. Innoscience’s device design and performance set the worldwide standard for GaN. The company's GaN are used in multiple low, medium, and high voltage applications, with GaN process nodes covering 15V to 1200V. With over 800 patents granted or pending, Innoscience’s products are known for reliability, performance, and functionality within the fields of consumer electronics, automotive electronics, data centers, renewable energy and industrial power. Innoscience creates a bright future for GaN. Please visit www.innoscience.com for more information.