AWS announces genAI Nova models, supercomputer plans in bid to catch up with rivals

The news: Amazon Web Services (AWS) unveiled Nova, a family of new generative AI (genAI) models, at its re:Invent conference this week.

The suite includes an image-generation model, a video-generation model, and four text-generation models called Micro, Lite, Pro, and Premier. All but Premier are available now on AWS.

AWS also announced updates for its Bedrock service to build out AI agents and catch hallucinations in models.

Multimedia models: This bombshell release, particularly the Nova Reel video generator, counters the slow and secretive approach OpenAI has taken with Sora.

• Reel can make videos up to 6 seconds long from either prompts or reference images. AWS said a version that can make 2-minute videos is coming soon.
• It lacks audio processing and struggles with counting and event causality—limitations common in current video generators.

Reel and Nova Canvas, AWS’ image generator, include digital watermarking on all outputs and guardrails against harmful content like misinformation, child sexual abuse material, and chemical or biological threats.

Hardware independence: The company is also building a supercomputer in partnership with Anthropic that will be powered by AWS’ Trainium hardware.

• This in-house chip development could reduce Amazon’s reliance on industry giant Nvidia.
• “Today, there’s really only one choice on the GPU side, and it’s just Nvidia. We think that customers would appreciate having multiple choices,” AWS CEO Matt Garman said, per The Wall Street Journal.

What’s next? AWS plans to release speech-to-speech and “any-to-any” models next year. That’s an ambitious plan considering none of the released Nova models can process or generate audio content.

• Breakthroughs in audio processing could help push forward its long-suffering Alexa revamp, which was recently delayed until next year even though Amazon announced it in 2023.
• However, if genAI audio capabilities are limited to AWS model development, we could see even longer Alexa delays.

Our take: This massive rollout signals an aggressive push to catch up with rivals like Google and Microsoft and validate AWS’ enormous AI spending.

Getting a pioneering any-to-any model out in the next year might stretch AWS teams thin and push Amazon to rely more heavily on partnerships with AI startups.

This article is part of EMARKETER’s client-only subscription Briefings—daily newsletters authored by industry analysts who are experts in marketing, advertising, media, and tech trends. To help you finish 2024 strong, and start 2025 off on the right foot, articles like this one—delivering the latest news and insights—are completely free through January 31, 2025. If you want to learn how to get insights like these delivered to your inbox every day, and get access to our data-driven forecasts, reports, and industry benchmarks, schedule a demo with our sales team.