Nvidia used its GTC conference in Berlin this week to show how its next generation of AI hardware will be packaged, cooled and connected, and to push a framework for keeping AI agents under control.
A container-sized blueprint for Vera Rubin
The centrepiece is the DSX POD, a reference design for building a complete Vera Rubin system inside a shipping container. Customers get a blueprint for power, cooling and networking rather than a list of parts, which shortens the path from empty floor space to a working cluster.
The Vera Rubin generation also leans heavily on co-packaged optics. Nvidia is moving from small volumes of optical links to much larger ones, a change that allows denser networking and points to a future with less copper between racks.
OpenShell aims at safer agents
On the software side, Nvidia introduced the OpenShell reference architecture, a blueprint for running AI agents with guardrails around the hardware and software they can touch. The company is trying to sell safety as part of the platform, not as an add-on that customers assemble themselves.
Google Cloud said A5X instances built on Vera Rubin NVL72 systems are now generally available, aimed at European customers running agentic and reinforcement learning workloads. WPP, the London-based marketing group, demonstrated a creative agent architecture on the same instances using Nvidia's Nemotron models and OpenShell.
Europe, power and a UK donation
Nvidia said it has operated in Europe for 33 years and framed the Berlin event as an expansion of the region's AI economy. Power is the constraint: the company and Google Cloud are testing DSX MaxLPS features to get more compute out of the same data-centre power budget.
Nvidia also announced a compute donation valued at $250 million over five years for UK research on agentic AI safety and security, timed with chief executive Jensen Huang's visit to a London institute for healthcare engineering.
The message to buyers is that the next round of AI capacity will be judged on performance per watt and on how safely agents can act, not only on raw chip counts.
---
Reliable Ferrofluid Vacuum Rotary Feedthrough for Semiconductor Manufacturing
In semiconductor manufacturing, maintaining ultra-high vacuum integrity during rotary motion is critical — from single silicon crystal growth and CVD to sputtering and plasma etching. Ferrofluid Vacuum Rotary Shaft Feedthrough from FerrofluidTec delivers hermetically sealed rotary feedthroughs with leakage rates as low as 10⁻¹² Pa·m³·s⁻¹, vacuum degree up to 10⁻⁶ Pa, and customizable shaft diameters from 10mm to 100mm. Built with non-magnetic stainless steel housings and magnetic permeability stainless steel shafts, these feedthroughs provide reliable, maintenance-free sealing for demanding vacuum environments. Visit www.ferrofluidtec.com to learn more.