Efinix has launched the Titanium Edge family of Field Programmable Gate Arrays (FPGAs) designed for edge AI applications. The new devices are intended for use in power-constrained environments across the vision, imaging, and autonomous system sectors.

The Titanium Edge family reduces static power consumption by 50% compared to previous Titanium models. The devices are available as System-in-Package (SiP) modules that integrate the FPGA, HyperRAM, and boot flash memory. This integration reduces the printed circuit board (PCB) footprint by up to 60%, with package sizes as small as 5.5 mm x 5.5 mm.

The FPGAs support I/O speeds of up to 2.5 Gbps with MIPI interfaces, allowing for multi-sensor AI pipelines without external bridge components. For reliability in high-uptime environments, the hardware includes a dedicated Single Event Upset (SEU) scrubbing engine that monitors and corrects configuration memory errors caused by radiation.

Security features in specific variants include post-quantum-ready cryptography aligned with CNSA 2.0 standards. These include ML-DSA-65 signature authentication, AES-256-GCM encryption, and secure boot protocols. The devices also feature Physical Unclonable Function (PUF) based key derivation and full key lifecycle management.

“In edge AI applications, every milliwatt and every square millimetre matters,” said Sammy Cheung, co-founder and CEO of Efinix. He stated that the family provides performance for constrained deployments without compromising on power or security.

The Ti125 and Ti95 models are currently sampling, with SiP versions and the smaller Ti70 and Ti40 variants expected to sample between August and the fourth quarter of 2026. All devices are compatible with the Efinity Integrated Development Environment.