Ceva's NeuPro NPUs Secure Multiple Licenses, Accelerating On‑Device Edge AI

Ceva’s NeuPro Wins Drive On‑Device AI Push

Strategic NPU Licenses Validate Scalable Edge AI Strategy

Ceva, Inc., headquartered in Rockville, Md., says 2025 marks a breakthrough year for its artificial intelligence licensing business after signing 10 NeuPro™ neural processing unit (NPU) agreements and reporting that AI contributes more than 20% of annual licensing revenue. The company says accelerating demand for optimized, power‑efficient neural processing architectures for edge and hybrid AI models is driving adoption of on‑device inference across diversified smart‑edge markets, positioning Ceva as a foundational enabler of what it calls Physical AI.

A standout development is a strategic NPU licensing agreement with a leading PC original equipment manufacturer, which selects Ceva’s high‑performance NeuPro NPUs as foundational IP for its next‑generation on‑device AI compute architecture. Ceva frames the deal as validation of its scalable NPU strategy across PC, automotive, advanced consumer and other compute‑intensive platforms, underlining a move by major OEMs to embed AI inference into end devices rather than rely solely on cloud processing.

Ceva’s 2025 NPU licenses span ultra‑low‑power to high‑performance inference for audio, vision, video and multi‑sensor workloads and cover consumer, automotive, industrial, PC and infrastructure markets. Earlier deals include a comprehensive NeuPro portfolio license with Microchip Technology and wins with ALi Corp and Nextchip, which expand Ceva’s footprint into consumer electronics, video platforms and automotive ADAS. The company expects six customers to have silicon back by the end of 2026, and several agreements involve existing high‑volume connectivity partners increasing content per device and long‑term royalty potential.

Automotive DSPs and Royalty Visibility

Complementing NPU momentum, Ceva signs multiple AI digital signal processor licensing agreements in 2025 across automotive, which the company says begin translating into measurable royalty ramps in 2026 and beyond. As customers move from AI evaluation to high‑volume deployment, Ceva anticipates clearer revenue visibility and upward pressure on market share in segments that require certified, power‑efficient inference engines.

Physical AI Positioning and Market Reach

Ceva emphasizes its role at the intersection of connectivity, sensing and on‑device inference, describing its IP as enabling Physical AI inside intelligent devices. By covering a spectrum of performance and power envelopes and securing agreements with chipmakers and OEMs, Ceva positions itself to capture growth as edge and hybrid AI architectures proliferate across smart‑edge applications.