Ambarella has launched the N1-665 system-on-chip and the Cooper Pro developer platform, designed to process large language and vision models locally on edge devices. The hardware supports up to 12 simultaneous 1080p30 video streams, targeting autonomous robotics and smart-city video applications.
The launch of the N1-665 provides a direct alternative to NVIDIA Jetson platforms for high-density video processing at the edge. By integrating a neural vector processor that handles 12 concurrent streams under 20 W, Ambarella is addressing the power-efficiency requirements of autonomous systems that cannot rely on cloud latency. This shift toward local VLM processing suggests a move away from centralized AI architectures in smart-city and industrial video deployments. As the ecosystem matures, watch for how the Cooper Foundry 2.1 software adoption rates compare to established proprietary AI stacks in the robotics sector.
Ambarella has been building out its Cooper ecosystem as a competitive alternative to NVIDIA's Jetson platform for edge AI workloads. At DTW IGNITE 2026, Nokia teamed up with Google Cloud to infuse AI into telecom operations, demonstrating how agentic AI is being deployed across infrastructure categories, though Ambarella's approach targets a different deployment model focused on local inference rather than cloud-orchestrated agents. The N1-665 positions Ambarella to capture robotics and smart-city deployments where power budgets and thermal constraints make cloud-dependent architectures impractical.
On the business side, Ambarella's strategy of offering an open software stack through Cooper Foundry 2.1 contrasts with NVIDIA's proprietary CUDA ecosystem. Nokia and Ericsson announced a collaboration to advance open standards and promote compatibility across their rApp portfolios, illustrating how multivendor interoperability is becoming a competitive differentiator in adjacent infrastructure markets. Ambarella's approach mirrors this philosophy by providing developers with a platform-agnostic toolchain that reduces switching costs compared to locked-in GPU programming environments.
From a technical standpoint, the N1-665's ability to handle 12 simultaneous 1080p30 streams under 20 W represents a significant power-efficiency claim for edge video processing. Ericsson introduced rApp as a Service on AWS, delivering AI-powered network optimization with 98% field-validated accuracy and 54% faster cell issue resolution, showing how AI inference at scale is being measured in production environments. For buyers evaluating the N1-665 against NVIDIA Jetson for autonomous robotics, the key differentiator will be whether Ambarella's physical edge AI hardware can sustain multi-stream throughput without the thermal throttling that affects GPU-based solutions in fanless enclosures.
Ambarella has launched the N1-665 SoC, a hardware solution designed for local large language and vision model processing on edge devices. By supporting 12 simultaneous 1080p30 video streams under 20W, it offers a power-efficient alternative to cloud-dependent architectures, specifically targeting autonomous robotics and smart-city security infrastructure that requires low-latency, local AI inference.
The N1-665 SoC is designed for edge AI, capable of processing 12 simultaneous 1080p30 video streams while running large language and vision models locally under a 20W power profile.
The hardware architecture supports multi-modal models including LLaVa-OneVision, Llama, and Gemma without requiring cloud reliance.
Cooper Pro is a developer platform that includes quad GMSL links, 10-Gb Ethernet, and 64-GB DRAM to support high-bandwidth video applications for the N1-665 SoC.
Ambarella offers an open, platform-agnostic toolchain through its Cooper Foundry 2.1 software suite, which contrasts with NVIDIA's proprietary CUDA ecosystem.
Add StreamingMeme as a preferred source on Google to see more of our streaming news at the top of your Search results.
Add as preferred source