NTT Unveils Low-power AI Inference Chip for Real-Time 4K Video Analysis at Upgrade 2025

At Upgrade 2025, NTT’s annual research and innovation summit held in San Francisco on April 9–10, Kazuhiro Gomi, President and CEO of NTT Research, took the stage to announce a new AI inference large-scale integration (LSI) chip that enables real-time processing of 4K video with low power consumption. The announcement aligned with the summit’s theme, “Innovation for the New Reality,” and marks a step forward in edge computing and AI video analysis.

“This chip runs AI inference on full 4K video without lowering the resolution,” Gomi said. “You can mount it on a drone flying 500 feet in the air and still detect human activity on the ground—something previously unachievable with conventional chips.”

The LSI is designed for use in edge devices and terminals where power constraints have traditionally limited the deployment of AI processing for ultra-high-definition (UHD) video.

Current AI inference systems often require the compression of ultra-high-definition video to enable real-time processing on power-limited devices. NTT’s new chip addresses this challenge by enabling high-resolution inference at 30 frames per second (fps) while consuming less than 20 watts of power.

One application involves drone-mounted cameras used in infrastructure inspections and public safety monitoring. With the new chip, drones can identify individuals or objects from altitudes up to 150 meters (492 feet)—the legal flight ceiling for drones in Japan. Previously, such tasks were restricted to lower altitudes due to the computational and energy demands of high-resolution inference, limiting their operational scope to around 30 meters (98 feet).

The chip’s efficiency is driven by an AI inference engine developed by NTT. The engine boosts computational performance through techniques such as interframe correlation and dynamic bit-precision control, enabling real-time execution of object detection algorithms like YOLOv3 without requiring GPU-level power consumption.

To accomplish full 4K processing within energy constraints, the chip divides the high-resolution video into segments and applies algorithms that ensure object continuity across segment boundaries. The system maintains detection accuracy while reducing workload.

Privacy was also a key factor in the chip’s design. NTT Research has integrated encryption technology that can obscure sensitive details like faces at the point of capture. Authorized users can later decrypt this information when necessary, preserving privacy while still enabling post-event analysis.

Commercial sales of the chip are scheduled to begin by the end of the year through NTT Innovative Devices, the company’s semiconductor division.

The chip was one of the key highlights announced at Upgrade 2025, alongside other initiatives focused on advancing next-generation infrastructure. NTT researchers are also exploring the integration of this LSI into the Innovative Optical and Wireless Network (IOWN) Initiative. The goal is to leverage the high-speed, low-latency capabilities of the IOWN All-Photonics Network to support scalable, energy-efficient data-centric infrastructure (DCI).

Additionally, NTT is collaborating with NTT DATA to explore how the chip’s capabilities could be enhanced using Attribute-Based Encryption (ABE) technologies. ABE provides fine-grained access control, enabling secure data sharing that aligns with flexible policy requirements and existing system architectures.

In edge environments, the chip opens new possibilities for real-time AI across industries such as transportation, utilities, and public safety.

With this new AI inference chip, NTT Research is pushing the boundaries of edge computing and real-time video analysis—offering a powerful tool for next-generation imaging applications.

Filed in General. Read more about AI (Artificial Intelligence), Encryption, Ntt, Processors, Research and Semiconductors.