Amazon [AP=Yonhap News file photo]

Amazon, the world's largest cloud company, is overhauling its artificial intelligence (AI) strategy by scrapping its proprietary AI model, 'Nova'.

Amazon has disbanded its existing Nova development organization and is refocusing its resources on the 'Frontier Model Research' (FMR) initiative led by newly recruited Vice President Pieter Abbeel, according to multiple sources cited by the U.S. business media outlet Business Insider on the 29th (local time).

Amazon has initiated a phased decommissioning process for models it had been developing under the Nova line, including the high-performance models 'Premier' and 'Omni', the video model 'Reel', and the image model 'Canvas'.

Vice President Abbeel, who is in charge of developing the new AI models, is a professor at the University of California, Berkeley (UC Berkeley) and a co-founder of the robot AI startup Covariant.

He joined Amazon after the company acquired Covariant in 2024.

The FMR team, led by Vice President Abbeel, is currently developing a new foundation model to be unveiled at Amazon's annual technology conference, 're:Invent', later this year. The development of this model has emerged as the company's top priority.

Consequently, it appears that Amazon is replacing the existing Nova models by developing a new one from scratch.

Reuters recently reported that Amazon had previously laid off some employees in its Artificial General Intelligence (AGI) division.

An Amazon spokesperson stated, "It remains unchanged that the development of AI models is one of the most important tasks we are pursuing."

However, Amazon has retained certain Nova-line assets, including the lightweight models 'Nova 2 Sonic' and 'Nova 2 Lite', the model-building service 'Nova Forge', and the AI agent technology 'Nova Act'.

There is also a possibility that the newly developed model will inherit the 'Nova' brand in the future.

Amazon's 'model overhaul' is considered similar to the strategy adopted last year by its AI industry rival, Meta.

When its open-source model 'Llama' failed to meet expectations, Meta recruited Alexandr Wang, founder of the startup 'Scale AI', as the head of the Meta Scale Lab (MSL) and introduced a new AI model called 'Muse'.