Nvidia Vera Rubin goes live on Oracle Cloud at gigawatt scale

Vera Rubin NVL72 systems and Vera CPUs are generally available on OCI, with a multi-gigawatt expansion planned.

On October 27, 2026, Nvidia and Oracle announced that Nvidia Vera Rubin NVL72 systems and Vera CPUs are now generally available on Oracle Cloud Infrastructure. The two companies also plan to expand OCI Superclusters with Vera Rubin to multi-gigawatt scale. Here is what was announced, what is still a plan, and why the CPU side matters for AI agents.

Key facts

On October 27, 2026, Nvidia and Oracle announced that Vera Rubin NVL72 systems and Nvidia Vera CPUs are generally available on Oracle Cloud Infrastructure (OCI).
The companies plan to expand OCI Superclusters with Vera Rubin NVL72 to multi-gigawatt scale. The release gives no specific gigawatt figure and no contract value.
OCI also plans to introduce HGX Rubin NVL8 systems, with no date given.
Each NVL72 rack connects 72 GPUs through NVLink, and OCI’s network can scale across hundreds of thousands of GPUs.
OpenAI is an early-access customer for Vera CPUs on OCI and says it is bringing them into production there.

What happened

Nvidia and Oracle published a joint announcement at Oracle AI World. The headline is availability: Vera Rubin NVL72 systems and Vera CPUs can now be used on Oracle Cloud Infrastructure. They build on the GB200 and GB300 NVL72 systems that OCI already offers.

The second part is a plan. The companies say they intend to grow OCI Superclusters built on Vera Rubin NVL72 to multi-gigawatt scale. A Supercluster combines NVL72 racks, Oracle Acceleron networking, Nvidia Spectrum-X Ethernet with Spectrum-6 switches, BlueField-4 DPUs, and power and cooling designed for gigawatt-class sites. The release does not say when each stage of the expansion will arrive.

OCI also plans to add HGX Rubin NVL8 systems: eight Rubin GPUs linked by NVLink in an x86 server with two 64-core Intel Xeon 6 CPUs, 4 TB of DDR5 memory and 6.4 Tbps of RDMA networking. No timeline was given.

The release says OpenAI trained and served GPT-6 Astra in Abilene on an OCI Supercluster with more than 100,000 Nvidia Blackwell GPUs. It also says OpenAI is among the early-access customers testing Vera CPUs on OCI and is bringing them into production. Tom’s Hardware, which covered the launch separately, describes it as a step for Nvidia beyond selling CPUs only inside its rack systems, since the Vera CPU is now offered as its own instance type at a major cloud provider. Performance claims in the vendor material come from Nvidia and were not independently verified in the coverage we reviewed.

Why it matters

Most AI headlines are about models, but models run on hardware, and the hardware is mostly rented from a few clouds. When a cloud makes a new generation of chips generally available, it sets the pace for what the AI products you use can do and how fast they can grow.

The Vera CPU is the detail worth noticing. According to the release, it is aimed at the work AI agents do between model calls: running tools, executing code and retrieving data. That is the part of an agent that browses, runs a script or opens a file. Oracle also points to hardware-based isolation for untrusted, AI-generated code, which matters as agents run more code on their own.

For everyday users nothing changes today: there is no new app to open. The practical effect shows up over months, as providers get more capacity for training and for serving agents. For anyone following the business side, the announcement is one more sign that AI spending is moving from announcements to physical capacity such as racks, networks and power. It is a fact about infrastructure, not advice about any company’s shares.

How to try it

This is an infrastructure story, so the useful skill is turning dense technical news into something you can act on. The Try it prompt, Explain simply, rewrites any complicated text in plain language.

1. Open the prompt and paste it into ChatGPT, Claude or Gemini. 2. Where it asks for the text, paste a paragraph from the announcement, for example the part about what Vera CPUs are for. 3. Where it asks for your level, write something like “I use AI tools at work but I am not an engineer”. 4. Ask for three things: a plain summary, the one term you should look up, and one question to ask your IT team or cloud provider.

Check the answer against the original. If the reply states a number, a date or a customer that is not in the text you pasted, ask the assistant to point to the sentence it came from. Treat anything it cannot point to as unverified.

Try it

Explain a topic simply

Frequently asked questions

What is Nvidia Vera Rubin?

It is Nvidia's current data center platform, pairing Rubin GPUs with Vera CPUs. The NVL72 version links 72 GPUs in one rack through NVLink so they work as a single large system for training and running AI models.

What does multi-gigawatt scale mean?

It describes the electrical power of the data centers involved, measured in gigawatts, and signals very large capacity. In this announcement it is a plan: the release gives no exact gigawatt number, no schedule and no contract value.

Is this available to everyone?

Vera Rubin NVL72 and Vera CPUs are described as generally available on OCI, so Oracle Cloud customers can use them. HGX Rubin NVL8 is only planned. Regular chatbot users do not need to do anything.

Sources