
Amazon is reportedly pulling back on most of its Amazon Nova AI models, the ones it built to compete with ChatGPT and Gemini. Instead of upgrading them, the company is keeping them barely running and starting over with a small new research team.
The Gist
- Amazon is scaling back several of its Amazon Nova models, including its top text model and its image and video generators.
- Those models will keep working for existing customers but are no longer being actively improved.
- A new team is being built to design a fresh, more powerful model, with a first result expected in late 2026.
Have ChatGPT Recap This Article
ChatGPTMeet Nova, Amazon’s own answer to ChatGPT
You have probably never opened an app called Nova, and that is part of the story. Amazon Nova is Amazon’s own family of AI models, its in-house answer to ChatGPT, Gemini and Claude. If you want a refresher on what an AI model actually is under the hood, that plain-English explainer covers it.
Unlike ChatGPT, Nova was never really built to be a consumer app you chat with directly. It mostly lived behind the scenes, offered to businesses through Amazon’s cloud service so they could build their own tools on top of it. So when Amazon changes course on Nova, you will not see an app disappear from your phone. What changes is the pecking order of who is really competing at the top of AI.

What Amazon is quietly switching off
The reporting names specific models. Amazon is scaling back Nova Premier, its most capable text model, along with Nova Omni, plus Reel, its video generator, and Canvas, its image generator. These are the tools that would have gone head to head with the AI that writes, draws and makes short clips for millions of people.
The models are not being deleted. Insiders describe them as staying in a keep the lights on mode, meaning they keep running for the businesses already using them, but nobody is actively making them better anymore. In AI, standing still is the same as falling behind, because rivals ship upgrades every few weeks. Some of these tools competed in the same space as the AI video tools now good enough to make a full movie.
Amazon has not officially confirmed the pullback. A company spokesperson pushed back gently, saying AI models remain one of Amazon’s most important projects. That framing matters, because the company is not quitting AI. It is changing who builds it and how.
Keep learning on AI Noobies:
- Apple Smart Glasses Delayed Over Privacy Fears
- RAM Shortage Is Making Your Next Phone Cost More
- Gemini Daily Brief Reads Your Inbox Each Morning
Why even a giant like Amazon hit a wall
Here is the surprising part. Amazon is one of the richest companies on earth, and it still could not keep its Amazon Nova models at the front of the pack. Building a top AI model takes enormous computing power, huge amounts of data, and some of the most expensive talent in tech, and the leaders keep raising the bar.
So Amazon is regrouping around a new research group, a Frontier team focused on building one strong foundation model from the ground up. A foundation model is the big, general-purpose brain that everything else gets built on top of. The team is led by Pieter Abbeel, a well-known AI researcher who joined Amazon when it bought his robotics startup, and a first model is expected at the company’s big conference in late 2026.
The wider debate this feeds is about who gets to win AI at all. When even Amazon steps back to regroup, it pushes more of the conversation toward a handful of front-runners like OpenAI, Google and Anthropic. That is the same concentration you can already feel when Google keeps making Gemini cheaper and faster while smaller efforts fade.
What Amazon’s retreat means for the apps you use
For your day to day, nothing breaks tomorrow. If you use Alexa, an Amazon app, or a service built on Amazon’s cloud, it keeps working exactly as before. The change is slower and quieter than that.
What it really tells you is where to place your attention. The AI tools most worth learning are the ones getting better every month, and right now those are the big consumer names, not Amazon’s paused models. If you have been meaning to test one seriously, pick the tool that keeps shipping updates, because that is where your practice will still be useful a year from now.
The story of Amazon Nova is a useful reminder for beginners. The AI race is moving so fast that even a trillion-dollar company can fall behind and have to start over, so you do not need to feel behind for still learning the basics.
Stay tuned on AI Noobies.



