Amazon designs custom AI chips for Echo and Fire TV devices

Amazon has gone full vertical. The company unveiled custom-designed AZ3 and AZ3 Pro chips built specifically for its Echo, Fire TV, and future consumer devices, each packing dedicated AI Accelerators meant to handle artificial intelligence workloads directly on the hardware rather than bouncing everything to the cloud.

Instead of sending every voice command to a distant data center and waiting for a response, these chips process AI tasks locally. The result, according to Amazon, is a wake-word detection improvement of over 50% on the AZ3-equipped Echo Dot Max, along with significantly better background noise filtering.

What Amazon actually built

The new silicon comes in two flavors. The AZ3 powers devices like the Echo Dot Max, while the AZ3 Pro targets more demanding hardware like the Echo Studio. Both chips feature what Amazon calls dedicated AI Accelerators, purpose-built processing units designed to run machine learning models at the edge.

These chips underpin a platform Amazon is calling Omnisense, which handles audio processing and environmental sensing. Advanced microphone arrays working in tandem with the on-device AI enable improved speech recognition even in noisy environments.

The broader device lineup announced on September 30, 2025, includes the Echo Dot Max, Echo Studio, Echo Show 8, and Echo Show 11. All ship with Alexa+ Early Access, Amazon’s next-generation AI assistant that leans heavily on personalized and proactive interactions. The Echo Dot Max and Echo Studio started shipping around October 29-30, 2025, with the Echo Show models following on November 12, 2025.

Fire TV also gets pulled into the strategy. Amazon has been integrating Alexa+ across its home entertainment lineup, positioning these custom chips as the foundation for a unified smart home ecosystem.

A decade-long silicon strategy

This isn’t Amazon suddenly deciding to play chipmaker. The company has been quietly building its custom silicon capabilities since acquiring Annapurna Labs back in 2015. That Israeli chip design firm became the engine behind Amazon’s in-house hardware ambitions.

By 2018, reports surfaced that Amazon was developing AI-focused chips specifically aimed at making Alexa faster and more responsive. The AZ3 family represents the maturation of that effort, nearly a decade in the making.

Amazon’s cloud division, AWS, has already shipped multiple generations of custom processors: Graviton for general computing, Inferentia for AI inference, and Trainium for model training. The AZ3 chips bring that same philosophy down to the consumer device level.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.