In his first AWS re:invent since taking over as Amazon Web Services (AWS) CEO in June 2024, Matt Garman declared that there had “never been a more exciting time out there in the world to be a start-up” as he announced a slew of developments set to maximize the potential of AI. Also at the event, AWS announced Amazon Nova, a new generation of state-of-the-art foundation models (FMs) that deliver frontier intelligence and “industry leading price performance”, available exclusively in Amazon Bedrock.

AWS re:Invent 2024. Photo by Noah Berger/Getty Images for Amazon Web Services

Garman – who has been at AWS since 2006 and was involved in the initial launch of its services – was speaking during the keynote of AWS re:Invent, an annual conference hosted for the global cloud-computing community. It’s an event that continues to grow, he indicated, with almost 60,000 people attending in-person in Las Vegas and a further 400,000 watching online.

“At its core, re:Invent is a learning conference dedicated to builders and specifically to developers,” said Garman. “The way the whole AWS developer community has grown is incredible. We now have 600 user groups all around the world, spanning 120 different countries, and the entire AWS community is many millions all around the world. One of the great things about that community is we get feedback from you that goes directly into the products and informs what we build and announce here today.”

Every year since 2020, AWS has provided startups with nearly $1 billion in promotional credits, and Garman revealed that this would again be the case in the coming year. Meanwhile, having noted that “Generative AI has the potential to disrupt every single industry out there”, it was to be expected that much of the ensuing presentation was dedicated to computing instances able to facilitate intensive AI applications.

Specific announcements included the availability of Amazon EC2 Trainium 2 (TRM2) instances, which are said to offer the best price performance by AWS for GenAI; the forthcoming introduction, early next year, of the Nvidia P6 family of instances; and the EC2 TRM UltraServer, which connects 4 Trainium2 instances and offers over 83 petaflops of computer from a single Ultra node.

“We have more instances, more capabilities, and more compute than any other cloud,” said Garman.

On the topic of GenAI, Garman revealed details of Amazon Bedrock Model Distillation – which supports the creation of faster and more effective models – and Amazon Bedrock Multi-Agent Collaboration, which is designed to allow rapid build, deployment and orchestration of teams to work together on complex tasks.

Garman also discussed the new Amazon Aurora DSQL – a server-less distributed SQL (structured query language) database that promises “virtually unlimited scalability” – and Amazon S3 Tables storage, which is specifically designed for analytics workloads, resulting in up to 3x faster query throughput and up to 10x higher transactions per second compared to self-managed tables.

With the world appearing increasingly unstable at the precise moment the amount of data being generated is surging to unprecedented levels, it’s not surprising that Garman also took pains to stress AWS’ commitment to security, noting that it matters “in every single thing we do… Security is one of the reasons that so many customers are trusting AWS with our cloud workloads. It’s this core foundational layer that all of the rest of our services build on top of.”

Creative Content Generation: Bringing Concepts to Life 

In other AI news at re:Invent, Amazon introduced Amazon Nova, a new generation of foundation models (FMs) that includes Amazon Nova Reel, which generates studio-quality videos. Using Amazon Nova Reel, you can produce short videos through text prompts and images, control visual style and pacing, and generate professional-quality video content for marketing, advertising, and entertainment. Amazon Nova Reel outperforms existing models on human evaluations of video quality and video consistency.

The new Amazon Nova models will be available in Amazon Bedrock, and include Amazon Nova Micro (a very fast, text-to-text model with the lowest latency and lowest cost); Amazon Nova Lite (multimodal models that can process text, images and videos to generate text, including processing videos up to 30 minutes in length); Nova Pro (which AWS says sets new standards in multimodal intelligence and agentic workflows that require calling APIs and tools to complete complex workflows) and Amazon Nova Premier (still in “training” and expected to be the most capable model for complex reasoning tasks and for use as the best teacher for distilling custom models).

And for the visual image community AWS rolled out Amazon Nova Canvas and Nova Reel, a state-of-the-art image generation model producing studio-quality images with precise control over style and content, including rich editing features such as inpainting, outpainting, and background removal. Amazon Nova Canvas excels on human evaluations and key benchmarks such as text-to-image faithfulness evaluation with question answering (TIFA) and ImageReward. Nova Reel, meanwhile, is a state-of-the-art video generation model. Using Amazon Nova Reel, users can produce short videos through text prompts and images, control visual style and pacing, and generate professional-quality video content for marketing, advertising, and entertainment. AWS says Amazon Nova Reel outperforms existing models on human evaluations of video quality and video consistency ad all Amazon Nova models include built-in safety controls and creative content generation models include watermarking capabilities to promote responsible AI use.