Oracle unveiled a flood of new analytics and AI capabilities this week at its Oracle CloudWorld conference in Las Vegas. The festivities kicked off yesterday with expanded AWS and Google Cloud capabilities, updates to Fusion Data Intelligence, and new features in its HeatWave Lakehouse, GenAI, and AutoML solutions. The company then capped off its deluge of CloudWorld news by announcing a supercomputer with more than 130,000 Nvidia GPUs.
With 131,072 Nvidia Blackwell GPUs delivering a potential 2.4 zettaFLOPS of peak performance on the Oracle Cloud Infrastructure (OCI) supercluster, Oracle says it is now taking orders for the world’s largest AI supercomputer.
“We have one of the broadest AI infrastructure offerings and are supporting customers who are running some of the most demanding AI workloads in the cloud,” said Mahesh Thiagarajan, executive vice president of OCI, in a press release today.
Customers can order OCI superclusters powered by Nvidia H100 or H200 Tensor Core GPUs, or Blackwell GPUs. Oracle says superclusters powered by H100 GPUs can scale to 16,384 GPUs with up to 65 ExaFLOPS of performance and 13 bps of aggregate network throughput, while superclusters powered by H200 GPUs will scale to 65,536 GPUs with up to 260 ExaFLOPS of performance and 52 bps of aggregate network throughput.
Oracle also updated HeatWave, which it first launched in late 2020 as a fully managed MySQL database for its own Oracle Cloud Infrastructure (OCI) with the ability to host transactional and analytical workloads. Over the years, the company has added additional features, such as built-in machine learning. in 2022 and MySQL-based Lakehouse features in 2023.
Yesterday, Big Red unveiled several new HeatWave features for AWSincluding HeatWave Lakehouse Indexing, HeatWave GenAI, and HeatWave Autopilot.
The new HeatWave Lakehouse offering gives Oracle’s AWS customers the ability to leverage Amazon S3 as backend storage, while using Oracle engines to query structured, semi-structured, and unstructured data.
The new HeatWave GenAI capabilities will give AWS customers the ability to automate vector store creation and vector ingestion, as well as leverage large language models (LLMs) built into the database. Customers can also leverage Amazon Bedrock’s LLM models, Oracle said.
“This enables AWS users to build rich generative AI applications without the need for AI expertise, without complex manual integrations and troubleshooting, and without the security risks and costs of moving their data to separate services,” said Edward Screven, Oracle’s chief enterprise architect, in a press release.
HeatWave Autopilot support will allow AWS customers to build better indexes, while native JavaScript support will allow AWS users to natively run stored procedures and functions they wrote in JavaScript. In total, Oracle says the HeatWave extension on AWS will allow AWS customers to retire six separate AWS services.
Oracle also made updates to its HeatWave MySQL, HeatWave GenAI, and HeatWave Lakehouse offerings running on OCI.
HeatWave MySQL gets a new “hypergraph” optimizer to improve query plan cost optimization; integration with OCI Ops Insights to help administrators uncover performance issues, predict consumption, and plan capacity; and bulk ingestion, which will load data up to 5x faster.
Its HeatWave GenAI offering benefits from features such as OCR support, batch processing of LLM inference workloads, automatic vector store updates, support for 27 languages, and JavaScript support.
HeatWave Lakehouse now supports LLMs up to 4x larger than before; new topic modeling capabilities for things like sentiment analysis on social media data; improved data drift detection; and semi-supervised log anomaly detection.
Oracle also took advantage of its CloudWorld trade show to make several announcements around Fusion Data Intelligence, its range of analytics, ML and AI capabilities for its enterprise software products, such as ERP and CRM systems.
For example, Oracle has rolled out a People Leader Workbench application for Oracle Cloud HCM, designed to help businesses achieve their goals. Oracle Cloud SCM (supply chain management) users now have a new Supply Chain Command Center application that they can use to get recommendations on changing demand, supply, and market conditions.
Oracle owns Siebel, but is expanding its Fusion Data Intelligence suite to support the competing Salesforce CRM application. Specifically, Oracle supports merging data from customers’ Salesforce implementations with other systems within Fusion. Oracle also strengthened its core operational reporting capabilities and added new AI and ML capabilities to analytics across ERP, HCM, SCM, and CX suites.
Finally, Oracle rolled out a new developer assistant that uses GenAI to help configure Fusion Data Intelligence service, data, and applications.
Oracle also strengthened its partnerships with several cloud providers. As part of its partnership with AWS, both companies now support Oracle Exadata Database Service running on AWS. They also announced new “seamless” data transfer capabilities between the Oracle database and applications running on Amazon EC2, AWS analytics services, and even Bedrock.
“We’re seeing tremendous demand from customers who want to use multiple clouds,” said Larry Ellison, Oracle’s co-founder, chairman and CTO, in a press release. Ellison gave a speech at OpenWorld yesterday, which you can see here.
Oracle is also making progress with its software running in Google Cloud data centers. Specifically, Oracle announced that Exadata, its Oracle Autonomous Database, and its Oracle Standalone recovery service without data loss running on OCI in four Google Cloud data centers including .S. East (Ashburn), US West (Salt Lake City), UK South (London), and Germany Central (Frankfurt).
“This new service combines all the benefits of OCI database services with Google Cloud services for a seamless multicloud experience, something that was unthinkable in the cloud space just a few years ago,” said Karan Batta, senior vice president, OCI, in a press release.
Related articles:
Oracle Brings Autonomous Database to Microsoft Azure
Oracle offers more on-premises options for hybrid cloud
Oracle Announces General Availability of MySQL HeatWave Lakehouse