Amazon Web Services (AWS) has announced several new technologies at its annual re:Invent conference, designed to enhance AI capabilities, streamline development processes, and improve cloud computing performance.   On stage at re:Invent 2024 in Las Vegas, Amazon CEO Andy Jassy said the company is launching Amazon Nova, a new family of multimodal generative AI models that can process text, image, and video inputs. These models will be available through Amazon Bedrock, enabling businesses to experiment with various AI models tailored to their specific applications.  “This is the future of how frontier models can be built and consumed, and we really look forward to giving this to you,” said Jassy. To improve the reliability of AI-generated content, AWS has announced Automated Reasoning checks to minimise factual errors. The company has also launched features for multi-agent collaboration and model distillation, the latter allowing businesses to transfer knowledge from large, advanced models to smaller, more efficient ones. Meanwhile, in his first keynote since taking up the role of AWS CEO, Matt Garman told the audience that in the chip space the company has launched new EC2 Trn2 instances, featuring AWS’s newest Trainium2 AI chip, which it says will offer 30-40 per cent better price performance than the current generation. “These are purpose built for the demanding workloads of cutting edge generative AI training and inference,” said Garman. Looking ahead, AWS revealed it is developing Trainium3, its first AI chip built on a 3-nanometer process node. This technology is expected to deliver unprecedented performance, energy efficiency, and density.  The new chip will enable businesses to build and deploy even larger AI models more rapidly, according to AWS. Trainium3-powered instances are slated for release in late 2025.  For development teams, new enhancements to Amazon Q Developer are set to help businesses modernise Windows .NET applications to Linux, transform VMware workloads to cloud-native architectures, and accelerate mainframe modernisation processes. AWS also refreshed its machine learning portfolio with the launch of the SageMaker Unified Studio, designed to simplify data discovery and access across organisations. This platform integrates AWS’s advanced analytics, machine learning (ML), and AI tools, allowing businesses to address diverse data challenges effectively.   Another new offering, the SageMaker Lakehouse, bridges the gap between data lakes, warehouses, operational databases, and enterprise applications, enabling seamless integration and actionable insights across all data sources, according to AWS.   With the exponential growth in data stored on Amazon S3, finding specific objects has become increasingly complex. AWS introduced Amazon S3 Metadata, a tool that captures queryable metadata in near real-time. By storing this metadata in S3 Tables, businesses can accelerate analytics and streamline data discovery across expansive data lakes, according to AWS.   Cathy O’Sullivan travelled to re:Invent 2024 as a guest of AWS.