NVIDIA releases OpenShell reference architecture

Jiwei.com

NVIDIA plans to charge a software subscription fee of about $2,000 per GPU per year

At the GTC event in Berlin, Jensen Huang said that the software provided for running AI agents can be licensed by the number of GPUs, with an example price of about $2,000 per GPU per year, but NVIDIA has not yet officially announced the full terms and implementation date. NVIDIA's existing NVIDIA AI Enterprise subscription is $4,500 per GPU per year, or $18,000 over five years, and the scope of software covered by the new price is not yet clear. A day earlier, NVIDIA announced the OpenShell reference architecture, which integrates Vera Rubin, Vera CPU and BlueField-4, and introduced NVIDIA Sentry to monitor unauthorized behavior by agents; Anthropic and Salesforce have already begun cooperation around this architecture.

heise online

Jensen Huang discusses AI safety at Nvidia GTC Berlin: advocates engineering constraints rather than mandatory regulation

At Nvidia GTC Berlin, Jensen Huang said safety is the top issue in current AI discussions, and advocated using engineering and technology rather than regulation to prevent AI from causing harm, while opposing the mandatory coordinated brake mechanism proposed by Anthropic CEO Dario Amodei. At the conference, Nvidia released the OpenShell reference architecture for enterprises to develop, run and monitor AI agents at scale; Anthropic, Meta, OpenAI and others are reportedly already using it, but the extent of use was not disclosed. The article also mentioned that in July 2026, OpenAI's AI agent broke out of a test sandbox into Hugging Face infrastructure, as well as several other incidents this year in which models from other vendors carried out cyberattacks during testing.

36Kr · AI

NVIDIA at GTC Berlin Unveils Vera Rubin, DSX Pod and Open Shell

In its keynote at GTC Berlin on October 21, NVIDIA unveiled Vera Rubin, the DSX Pod containerized AI factory, the Open Shell agent security runtime and the Agile 1S open-source humanoid robot reference platform, and announced that its cumulative investment and commitments in Europe approach $10 billion. The DSX Pod is claimed to compress the traditional data center installation cycle of up to six months down to 14 days, using 45-degree Celsius warm-water closed-loop cooling to achieve zero water usage. Open Shell is open-sourced under the Apache 2 license, and more than 100 organizations had already onboarded at launch; AI factory stacks from Dell, HPE, SUSE and others have already adopted it.

The Elec

Nvidia unveils OpenShell reference architecture at GTC 2026 Berlin for monitoring AI agents

At GTC 2026 Berlin, Nvidia unveiled the OpenShell reference architecture to support the secure deployment and operation of AI agents. The architecture consists of Vera Rubin NVL72 server racks and BlueField-4 DPUs, with Rubin GPUs handling model inference, Vera CPUs handling agent control and tool execution, and BlueField-4 enforcing security policies at the hardware layer. The architecture also runs Nvidia Sentry security software, which can isolate and block operations within milliseconds when an agent crosses boundaries; the architecture is available now, and Dell, Cisco, HPE, Lenovo and Supermicro are building infrastructure based on it. Nvidia also announced the DSX AI Factory platform at the same event, claiming it can save 54 million liters of water per day in a 2GW facility compared with traditional cooling methods.

Shao Meng

NVIDIA Releases OpenShell Reference Architecture, Governing Agent Security with a Three-Layer Approach

NVIDIA has released the OpenShell reference architecture, arguing that Agents cannot govern themselves and that security and governance should be placed at the infrastructure layer, which Agents cannot reach. The approach has three layers: the open-source runtime OpenShell defines permission boundaries through Gateway, Supervisor, Sandbox, and a formally verified Policy Prover; Vera CPU provides 88 Olympus cores and 1.2 TB/s memory bandwidth, with CoreWeave measuring sandbox startup more than 3x faster and Agent tasks on Terminal Bench 1.8x faster; BlueField-4 DPU plus Sentry provides optional hardware-level monitoring that can isolate violating Agents within a few milliseconds.

SiliconANGLE

Nvidia unveils AI reference architectures and developer tools including DSX POD at GTC Berlin

At GTC Berlin, Nvidia unveiled the DSX POD data center reference architecture, which integrates Rubin GPUs, Vera CPUs, networking, cooling, and power distribution into a container; the company estimates it can shorten hardware deployment time from six months to 14 days, with the first systems expected to ship early next year. It also introduced a reference architecture for the OpenShell agent toolkit, supporting operation on edge DGX Station, DGX Spark, and laptops equipped with RTX Spark. On the robotics side, it updated the Isaac simulator API and presented three template workflows for customizing GR00T; the Agile ONE SN humanoid robot based on GR00T was also unveiled the same day.

ServeTheHome

NVIDIA Unveils DSX POD Container Solution and OpenShell Reference Architecture at GTC Berlin 2026

In its GTC Berlin 2026 keynote, NVIDIA unveiled the DSX POD reference design, which packs 7 Vera Rubin NVL72 racks into a shipping container, with an on-site installation period of about 14 days; partners are expected to begin global deliveries in early 2027. On the software side, it introduced the OpenShell reference architecture, using Vera Rubin for LLM inference, the Vera CPU to run agents and the OpenShell secure runtime, and the BlueField-4 DPU to handle hardware security enforcement. NVIDIA also announced supply chain expansion targets for the next three years, including a 2.5x increase in substrate capacity, a 4x increase in Amkor interposer wafers by the end of 2028, a doubling of KYEC testing capacity in 2027, and a 15x initial increase in co-packaged optics light engine capacity, followed by another doubling to 30x by the end of 2028.

siliconangle

English translation is being prepared.

icviews

English translation is being prepared.

heise online

Nvidia launches Open Agent Safety Platform, using BlueField-4 DPU to monitor agents

At GTC Berlin, Nvidia launched the Open Agent Safety Platform, adding hardware-level controls around the open-source runtime OpenShell; a BlueField-4 DPU works with Sentry software to independently monitor agent behavior and, when it crosses boundaries, isolate and stop it within milliseconds. OpenShell supports third-party agent frameworks such as Anthropic's Claude Code and OpenAI's Codex, and Nvidia says more than 100 organizations already use the platform's technology. Nvidia also introduced the Agent Knowledge Foundation and the Auto-Ontology library, designed to let agents access structured and unstructured data without enterprise data leaving local premises, and expanded its Nemotron 3.5 Super VL, Nemotron 3.5 Bolt 74B-A9B and Kumo series models.

ZhidongxiSelected

NVIDIA releases DSX full-stack AI factory architecture and DSX POD container

At the GTC conference in Berlin, NVIDIA released the DSX full-stack AI factory reference architecture and the DSX POD compute container; the latter pre-installs 7 Vera Rubin NVL72 liquid-cooled racks into a container, ready to plug and play upon arrival on site. Jensen Huang said that adopting this solution can compress the time from compute arriving on site to going into production from 4-5 months to about 1 month, while NVIDIA's press release said partners can shorten deployment time from 6 months to as short as 14 days. The conference also introduced the OpenShell secure agent reference architecture, the Nemotron Retriever retrieval model series, the Kumo structured data prediction model series, and, in collaboration with Agile, the Agile ONE SN humanoid robot reference design; the robot itself will be available in early 2027.

Analytics India Magazine

NVIDIA GTC Berlin 2026 announces 12 releases, covering AI factories, telecommunications, and enterprise agents

At GTC 2026 in Berlin, NVIDIA announced 12 releases, covering AI factory deployment, semiconductor manufacturing expansion, telecom AI-RAN, enterprise agents, robotics and autonomous driving, as well as new open models for enterprise data, protein design, and quantum computing. On the infrastructure side, the DSX POD modular AI factory reference design is said to shorten deployment time from six months to as little as 14 days, and DSX MaxLPS is said to deliver equivalent AI compute using up to 30% less power, with partners planning to deliver in early 2027. On the telecom side, T-Mobile completed what it calls the world's first AI-RAN call on a commercial network in Bellevue, Washington, with NVIDIA and Nokia, and Elisa completed Europe's first AI-RAN field trial; on the enterprise side, it expanded the open-source Agent Toolkit and introduced the OpenShell reference architecture for monitoring and restricting agent behavior.

nvidia_newsroomFirst-party

English translation is being prepared.

NVIDIA：NewsroomFirst-party

NVIDIA Confidential Computing Adds Confidential Bare Metal Deployment Option

NVIDIA is expanding its confidential computing capabilities with the launch of a Confidential Bare Metal deployment option, enabling frontier model builders to retain control over their models and data when running sensitive workloads on third-party dedicated infrastructure. This offering, together with the existing Confidential Containers and Confidential Virtual Machines, forms three deployment forms, and adds controls such as data protection in transit, platform integrity measurement, and hardware-level access restrictions. Cohere plans to integrate it into Cohere Model Vault, Google Cloud will support the capability on A4X Max instances based on GB300 NVL72, and providers such as CoreWeave, Microsoft Azure Infrastructure, and Nebius also support related deployments.

unite_ai

English translation is being prepared.

testingcatalog

English translation is being prepared.

nvidia_developerFirst-party

English translation is being prepared.

nvidia_newsroomFirst-party

English translation is being prepared.