The future of AI extends beyond centralized cloud data centers to locations throughout your organization, where high performance, ultra-low latency, and flexible deployment options drive competitive advantage. As part of Oracle’s distributed cloud strategy, we’re excited to announce the launch of Oracle Compute Cloud@Customer with NVIDIA GPU configurations (orderable today) and Oracle Private Cloud Appliance with GPU configurations (orderable in March), featuring NVIDIA L40S GPUs to bring enterprise-grade AI, graphics, and high-performance computing (HPC) capabilities directly into your location of choice—including your own data center.

The combination of Oracle distributed cloud platforms with NVIDIA GPUs enables organizations to run low-latency generative AI inferencing, large language model (LLM) fine-tuning, AI video analysis, and real-time digital twin simulations wherever needed. By deploying GPU-accelerated cloud capabilities throughout your enterprise, you get performance, flexibility, and scalability right where it’s needed—while also maintaining control of sensitive data to help you address your data residency and sovereignty requirements and gaining the benefits of cloud automation and economics with Compute Cloud@Customer.

Oracle Compute Cloud@Customer – Enabling AI to Run Anywhere

Oracle Compute Cloud@Customer is a fully managed hybrid cloud solution that lets you use Oracle Cloud Infrastructure (OCI) services including compute, storage, networking, and OCI Kubernetes Engine (OKE) in your data centers. It offers the same APIs, software developer kits (SDKs), and operational model as OCI, providing you with seamless workload portability and choice over where to run applications using cloud resources. Oracle Compute Cloud@Customer lets you use the latest high-performance infrastructure and cloud services with a cost-efficient pay-as-you-go pricing model.

For customers who have operational requirements to use similar capabilities with an on-premises, purchase-based financial model, we also offer Oracle Private Cloud Appliance, which uses the same hardware and provides the same infrastructure services while being owned and managed by the customer instead of Oracle.

With our new announcement, you’re able to add NVIDIA GPUs to both Compute Cloud@Customer and Private Cloud Appliance with the following key features:

• Independent scaling of GPUs, compute, and storage: up to 48 L40S NVIDIA GPUs, 6,624 OCPUs with 80.4 TB of memory, and a mix of up to 3.65 PB of high-capacity storage and 1.2 PB of high-performance storage.
• Powerful GPU VMs: up to four NVIDIA L40S GPUs, 108 Intel Xeon 8480+ CPU cores, 800-GB DDR5 memory, and 400 Gbps network bandwidth for the most demanding workloads
• Ultra-fast network connectivity: 800-Gbps data center connectivity that can directly connect Exadata Cloud@Customer or Exadata Database Machine to combine the power of GPUs with Oracle Database 23ai’s integrated AI Vector Search.
• Built-in OKE for simplified container management

Powering the Next Generation of AI With NVIDIA L40S GPUs

Built on the NVIDIA Ada Lovelace architecture, the NVIDIA L40S GPU is a multi-purpose GPU designed to deliver incredible performance for AI-intensive workloads, HPC, and graphics-rich applications. Each NVIDIA L40S GPU includes:

• Independent scaling of GPUs, compute, and storage: up to 48 L40S NVIDIA GPUs, 6,624 OCPUs with 80.4 TB of memory, and a mix of up to 3.65 PB of high-capacity storage and 1.2 PB of high-performance storage.
• Third-generation RT Cores and NVIDIA DLSS 3 for accelerated, AI-enhanced graphics performance

The L40S GPU can deliver up to 1.7 times the performance of an NVIDIA A100 GPU for AI use cases and includes best-in-class graphics capabilities, making it ideal for customers looking to expand capacity for AI or run mixed workloads. With tailored capabilities for AI inference, graphics, digital twins, and real-time 4K streaming, the L40S GPU opens up new opportunities for enterprises to innovate and scale.

“The addition of NVIDIA L40S GPUs to Oracle Compute Cloud@Customer opens up new opportunities for organizations worldwide. We can now meet customers’ most demanding enterprise-grade workloads in areas such as GenAI, graphics and high-performance computing—with the convenience of their own data centers. Organizations can run low-latency generative AI inferencing, LLM fine-tuning, and real-time digital twin simulations while maintaining control of sensitive data to help meet data residency and sovereignty requirements—and gain the benefits of cloud automation and economics.” — Matt Leonard, Vice President, OCI Edge Cloud Product Management

“With Oracle’s Compute Cloud@Customer offering, customers can run AI and graphics workloads at scale with up to 48 L40S GPUs, while maintaining control of their data and providing low-latency access to other data sources and consumers. The NVIDIA L40S GPUs’ added support for FP8 data types works in combination with the NVIDIA AI Enterprise and NVIDIA Omniverse platforms to enable the latest generative AI innovations to be deployed at an organization’s edge across a range of industries, including financial services, manufacturing, healthcare and more,” Irfan Ali, Global Head of Edge Solutions Sales, NVIDIA.

Real-World Impact

Accelerating Insights with Oracle Database 23ai

Oracle Database 23ai integrates AI vector data types and search capabilities directly into the industry’s most popular relational database. It helps enable AI-driven insights without requiring multiple isolated databases or data movement, helping increase performance and data security. By accelerating AI Vector Search on an Exadata Cloud@Customer and pairing it with Oracle Compute Cloud@Customer with NVIDIA L40S GPUs, you can deploy a complete distributed cloud solution in your data center to optimize your full compute, storage, networking, and data stack. Organizations rely on Exadata Cloud@Customer to run Oracle Database, and they can now rely on Compute Cloud@Customer for the application tier and supporting workloads.

Fraud Detection in Banking and Finance

By deploying AI-driven fraud detection systems on Oracle Compute Cloud@Customer or Oracle Private Cloud Appliance, banks can analyze millions of transactions daily with extreme speed and precision. These systems can be used to identify anomalous transaction patterns in real-time, helping reduce both response times and false positives. Securely storing and processing sensitive financial data on Exadata Cloud@Customer on premises can help financial institutions address stringent regulatory requirements while improving fraud detection accuracy. They can also integrate frameworks, such as NVIDIA Morpheus, as part of advanced solutions to further optimize AI-powered fraud detection workflows if needed.

Smart Factory Optimization with Digital Twins

By using Oracle Compute Cloud@Customer or Oracle Private Cloud Appliance, manufacturers can develop high-fidelity digital twins of their production lines. These digital replicas help enable real-time monitoring, predictive maintenance, and workflow testing without disrupting actual operations. Engineers can simulate new processes, optimize performance, and reduce downtime, driving significant cost savings and accelerating innovation.

Take the next step with Oracle Distributed Cloud GPU Resources

Oracle Compute Cloud@Customer and Oracle Private Cloud Appliance with GPU Expansion leveraging NVIDIA L40S GPUs are both designed to enable an era of AI-driven transformation—wherever you need it. Whether safeguarding financial systems, optimizing industrial operations, or delivering real-time insights at the edge, Oracle’s distributed cloud solutions empower enterprises to innovate faster, operate smarter, and maintain complete control over their data.

To learn more about Compute Cloud@Customer and Private Cloud Appliance with GPU Expansion, see Oracle Compute Cloud@Customer and Oracle Private Cloud Appliance.

All export, reexport, transfer and use of Compute Cloud@Customer, Private Cloud Appliance, and GPU Expansions will be subject to and in accordance with U.S. and applicable export control and economic sanction laws and regulations.

The preceding is intended to outline our general product direction. It is intended for information purposes only, and may not be incorporated into any contract. It is not a commitment to deliver any material, code, or functionality, and should not be relied upon in making purchasing decisions. The development, release, timing, and pricing of any features or functionality described for Oracle’s products may change and remains at the sole discretion of Oracle Corporation.