AWS reInvent: Everything Amazon’s announced, from new AI tools to LLM updates and more | TechCrunch

Amazon’s big AWS cloud event has begun, with a clear focus on the use of AI to keep its lead intact

Amazon Web Services is live in Las Vegas for its AWS re:Invent event, which kicked off November 27 and runs until December 1. We expect a rapid-fire stream of announcements and unveilings of recent things it’s been working on, particularly given the increased competition for cloud providers to meet the rise of AI with compelling offers for customers. AWS CEO Adam Selipsky’s keynote set the tone for the event, making it clear that AWS is in a position to defend its longstanding lead, and it’s deploying AI tools and services to continue being the top large cloud provider in the market.

We know you might not have time to watch the whole thing, so we’re taking that on for the next few days and will deliver quick hits of the biggest news as they are announced, all in an easy-to-digest, easy-to-skim list featuring on-the-ground reporting from editors and cloud experts Frederic Lardinois and Ron Miller, along with contributions from the wider TechCrunch team. Here we go!

Stay tuned for more product updates, announcements and analysis.

Wednesday, Nov. 29

Neptune Analytics

Another new tool is Neptune Analytics, which combines the best of both worlds: graph and vector databases. Ron reports that the new service helps customers analyze existing Neptune graph data or data lakes on top of S3 storage, taking advantage of vector search to find key insights. Read more.

Clean Rooms ML

Amazon’s launching a privacy-preserving service that lets AWS customers deploy “lookalike” AI models trained for one-off company-company collaborations called Clean Rooms ML, which Kyle writes is an offshoot of AWS’ existing Clean Rooms product and removes the need for AWS customers to share proprietary data with their outside partners to build, train and deploy AI models. Read more.

SageMaker HyperPod

From Amazon’s AWS cloud arm comes the SageMaker HyperPod, a new purpose-built service for training and fine-tuning large language models. Frederic spoke with Ankur Mehrotra, AWS’ general manager for SageMaker, ahead of today’s announcement, who said, “SageMaker HyperPod gives you the ability to create a distributed cluster with accelerated instances that’s optimized for disputed training. It gives you the tools to efficiently distribute models and data across your cluster — and that speeds up your training process.” Read more.

AWS Titan image generator

You read that right, Amazon is joining the ranks of other big techs in finally releasing its own image generator. Kyle reports that the Titan Image Generator is now available in preview for AWS customers and can create new images when given a text description or customize existing images. Read more.

Tuesday, Nov. 28

Amazon Q: An AI-powered chatbot

The big announcement for the day was Amazon Q, an AI-powered chatbot for AWS customers. During his keynote, AWS CEO Adam Selipsky described it as being able to “easily chat, generate content and take actions. It’s all informed by an understanding of your systems, your data repositories and your operations.” Kyle reports that Q is trained on 17 years’ worth of AWS knowledge and will go beyond just answering questions — it will also do things like understand the nuances of app workloads and suggest AWS solutions and products for apps that only run for a few seconds. Read more.

Guardrails for Amazon Bedrock

The new Guardrails for Amazon Bedrock tool lets companies define and limit the kinds of language a model can use. For example, define topics that are out of bounds for the model, so it simply doesn’t answer irrelevant questions, Ron writes. Read more.

New AWS Trainium chips for AI models

Amazon unveiled the latest generation of its chips for model training and inferencing (i.e. running trained models). Kyle writes that Amazon already discussed AWS Trainium2, designed to deliver up to 4x better performance and 2x better energy efficiency than the first-generation Trainium. The second chip was announced this morning, called the Graviton4, is intended for inferencing. The fourth generation in Amazon’s Graviton chip family (as implied by the “4” appended to “Graviton”), it’s distinct from Amazon’s other inferencing chip, Inferentia. Read more.

Amazon S3 Express One Zone

Amazon has a major update to its S3 object storage service called Amazon S3 Express One Zone, a new high-performance and low latency tier for S3. Frederic reports that One Zone will offer a major performance improvement for data-intensive applications, including AI/ML training, financial modeling and high-performance computing. Read more.

Three new serverless offerings

Amazon announced three new serverless offerings to make it easier to manage Aurora, ElastiCache and Redshift serverless services. Ron writes, “Because each of these options is serverless, it means that Amazon manages all of the hardware in the background, and delivers just the right amount of resources you need, scaling up when needed without IT having to deal with all of the back-end management work.” Read more.

Amazon One palm-scanning expands from stores to security

AWS lifted the lid on a new palm-scanning identity service that allows companies to authenticate people when entering physical premises. Paul reports that Amazon One Enterprise builds on the company’s existing Amazon One offering, which it debuted back in 2020 to enable biometric payments in Amazon’s own surveillance-powered cashierless stores. Visitors to Amazon Go stores can associate their payment card with their palm print, allowing them to enter the store and complete their transaction by hovering their hand over a scanner. Read more.

New thin client virtual desktop environment devices

Amazon launched new $195 devices that allow enterprise users to access virtual desktop environments, like Amazon WorkSpaces, over the internet. Sarah writes that the devices are housed in Fire TV Cube hardware — a decision Amazon made to leverage existing expertise from the arm of the retail giant that makes streaming media players. The company explained its decision to build new hardware came from customer feedback about wanting to lower IT spending by replacing desktops and laptops with less expensive hardware. Read more.