Oracle has released a smaller version of its Dedicated Region offering.
The company has also launched a version of the OCI Roving Edge Device featuring GPUs.
The new Oracle Cloud Infrastructure (OCI) Dedicated Region configuration - dubbed Dedicated Region25 - is scalable, starting at three racks, and can deployed in a few weeks at a customer's data center or on-premises.
Dedicated Region25 has a 75 percent smaller launch footprint than other Dedicated Region offerings and gives customers access to more than 150 OCI artificial intelligence (AI) and cloud services. The offering will be available sometime in 2025.
“Our priority is giving customers the choice and flexibility to leverage cloud services in the model that makes the most sense for their business,” said Mahesh Thiagarajan, EVP of Oracle Cloud.
“With OCI’s distributed cloud capabilities, we’re helping customers deploy a dedicated cloud in a small, scalable footprint, build applications with the best services across cloud providers, and deploy AI infrastructure anywhere they want."
Thiagarajan added that this flexibility "helps our customers address their unique needs and support their cloud investments in delivering significant business value.”
In other Oracle Edge updates, the company has released a new version of its OCI Roving Edge Device with a three-GPU option, optimized for AI inferencing at the Edge, now generally available.
According to the company, this will enable customers to "manage mission-critical data at the edge and run domain-specific LLMs or computer vision even in remote or disconnected locations." OCI Roving Edge Infrastructure consists of multiple configurations of ruggedized and high-performance devices.
The offering comes equipped with 56 cores (102 virtual CPUs), 512GB of RAM, and up to 123TB of storage.
This month saw defense technology company Anduril deploy its Lattice software platform to OCI and the OCI Roving Edge Infrastructure globally. Lattice via OCI will give warfighters real-time data and automated decision advantages, as well as dynamic machine tasking, and human-on-the-loop automation.
Oracle is also now taking orders for what it claims will be the largest AI supercomputer in the cloud and will have up to 131,072 Nvidia Blackwell GPUs offering 2.4 zettaflops of peak performance, although it is referring to FP4 format, rather than FP64 or even FP8.