Amazon unveils ‘Nova’ AI models, looking to make its mark in the generative AI revolution

LAS VEGAS — Amazon CEO Andy Jassy returned to the AWS re:Invent stage Tuesday morning, delivering one of the company’s biggest announcements of the day: a new set of AI foundation models called “Nova.”

Set to be available alongside popular third-party models via the company’s Bedrock service, the Nova models represent one of the biggest attempts yet by the tech giant to make its mark in AI. The company is seeking to overcome the perception that it was slow out of the gate in the early days of the generative AI revolution.

Jassy, the former AWS CEO, explained that Amazon developed the new models for internal use and decided to share them publicly.

“You guys want a lot. There’s a lot to do,” he said. “It’s one of the reasons why we have continued to work on our own frontier models, and those frontier models have made a tremendous amount of progress in the last four to five months. And we figured if we were finding value out of them, you would probably find value out of them.”

As detailed by Amazon, the Nova family includes these models:

• Amazon Nova Micro, a text-only model that delivers low-latency responses at low cost.
• Amazon Nova Lite, a low-cost multimodal model for processing image, video, and text.
• Amazon Nova Pro, a multimodal model that combines accuracy, speed, and cost for a wide range of tasks.
• Amazon Nova Premier, which Amazon describes as “the most capable of Amazon’s multimodal models for complex reasoning tasks and for use as the best teacher for distilling custom models.” Premier will be available in the first quarter of next year, according to Amazon.
• Amazon Nova Canva for image generation
• Amazon Nova Reel for video generation.

In addition, Jassy teased ahead to two additional models in the works for next year, a speech-to-speech model and an “any to any” model as part of the Nova initiative.

“This is really multimodal to multimodal,” he said. “So you’ll be able to input text, speech, images, video and output text, speech, images and video. This is the future of how frontier models can be built and consumed.”

It’s part of a broader effort by Amazon to give AWS customers new AI capabilities, while using AI to bring new momentum to its cloud business. Jassy said the company’s insight from its internal teams is that developers want a broad selection of AI models to choose from, and he expects this to continue.

“We’re going to give you the broadest and best functionality you can find anywhere,” Jassy said. That, he said, “is going to mean choice.”

With Nova and other initiatives, Amazon is looking to use AI to build on its position in the cloud market. The company leads rivals Microsoft and Google in the cloud infrastructure market with 31% market share as of the end of the third quarter, according to Synergy Research Group.

AWS posted $27.4 billion in revenue in the third quarter alone, with $10.4 billion in operating income, up 50% from a year ago — representing more than 60% of Amazon’s overall operating profits.

At re:Invent this week, the company is also highlighting progress in its Trainium and Inferentia AI chips, its Amazon Q AI assistant, and its partnership with Anthropic, maker of the Claude chatbot, among many other announcements.

It was the first re:Invent keynote led by AWS CEO Matt Garman, a longtime AWS executive, since he was named to lead the cloud division earlier this year. Jassy hadn’t been back on stage at the annual conference since he succeeded Amazon founder Jeff Bezos as Amazon’s CEO in 2021.