Amazon has started developing a custom artificial Intelligence chip that would power the upcoming Echo devices and also enhance the response time for its digital voice assistant Alexa, reports The Information.
The move brushes shoulders with rivals Apple and Google, who already have custom AI hardware at various scales. Performing AI tasks can be so computationally intensive and often need custom-made chips for the devices itself and also custom-designed servers for data centers, where artificial intelligence is developed and trained and deployed through the cloud.
What's the difference
With the development of AI chips, the reliance on the cloud will be reduced and the response speed to requests and data processing will be increased. It's still unclear when the shift would take place, but the reports suggest that Amazon has been trying to amplify its chipmaking abilities since it acquired Israeli chipmaker Annapurna back in 2015.
The company also acquired blink, a security camera manufacturer in December last year for an undisclosed sum. Blink started cramming in its own video hardware after it faced issues selling the chips alone. This put the company on Amazon's radar, and now it is under Amazon's ecosystem. But what difference does Blink makes to the e-commerce giant remains to be seen.
The new custom-chips will bring some major upgrades to Alexa's assistant capabilities. The new chips will enable Alexa to respond to requests faster than it currently does. But there are long-term improvements as well, such as controlling both hardware and software front for Alexa will give Amazon a much tighter control over the user experience.
Amazon could also see a big rise in its revenue by using its own chips. The company would be able to bring new AI-driven features to the voice assistant without relying on third-party chipmakers.
Both Apple and Google have already begun the similar shift. Apple developed its own chips for the latest generation iPhones, such as the graphics processor and power management unit. The company developed a new neural engine to power its A11 Bionic chip by handling on-device processing for machine learning to run features like the FaceID and ARKit apps.