Nvidia Corp. used its GTC Berlin 2026 keynote on October 21 to roll out a modular data-center design it says can compress deployment from six months to 14 days, while expanding its robotics platform and locking in billions of dollars in supply-chain capacity1,2,3.
Prefab data centers and supply-chain commitments
The centerpiece infrastructure product is DSX POD, a reference design for Vera Rubin-based AI factories built inside modular containers that are integrated and validated before shipping. Partners on the design include Dell, HPE, Lenovo, Schneider Electric, and Vertiv. Nvidia Corp. said European partners are adopting its DSX AI Factory platform to build facilities that use less cooling water and less power per unit of AI output.
On the supply side, Nvidia Corp. committed $1.3 billion to Amkor to increase advanced packaging capacity by more than 300% by the end of 2028. King Yuan Electronics plans new U.S. facilities backed by a $1.1 billion capacity-reservation advance from Nvidia Corp. Substrate agreements with Ibiden, Samsung Electro-Mechanics, and Unimicron Technology Corporation target a 2.5-fold capacity increase over three years. Infineon Technologies is advancing work with Nvidia Corp. on gallium nitride-based power solutions for 800 VDC AI data center architectures.
Robotics: Isaac expansion and new hardware
CEO Jensen Huang said that the "ChatGPT moment for robotics is fast approaching". Nvidia Corp. expanded Isaac, its open robotics platform, with new partners: LG Electronics will supply digital assets for virtual-environment simulation, and its subsidiary Bear Robotics will apply the Isaac GR00T model to its own robot platform. Germany's Agile Robots plans to ship Agile ONE SN, a humanoid reference platform equipped with Jetson Thor and GR00T, in early 2027. ABB Robotics, Kuka, and Universal Robots are also expanding their use of Nvidia Corp. technology. Volvo is adopting Nvidia Corp.'s Hyperion platform and Alpamayo software stack for automated driving.
Telecom, agents, and scientific AI
T-Mobile completed what Nvidia Corp. described as the first AI-radio access network call on a live commercial network.
For enterprise AI security, Nvidia Corp. introduced OpenShell, a reference architecture combining Vera Rubin infrastructure for inference, NVIDIA Vera CPU for agent tool use, and BlueField-4 DPUs running NVIDIA Sentry for in-silicon enforcement. The partner list for OpenShell spans Anthropic, Amazon Web Services, Cisco, CrowdStrike, Dassault Systèmes, Dell, Google Cloud, HPE, IBM, Meta, Microsoft, OpenAI, Palo Alto Networks, Perplexity AI, SAP, ServiceNow, Siemens, and SpaceX Super Intelligence.
In scientific computing, Nvidia Corp. and the University of Washington's Institute for Protein Design introduced RFD4-Proteina, an open AI model for protein design that Nvidia Corp. said generates designs about 13 times faster than RFdiffusion3 in a benchmark involving 200-amino-acid proteins. Nvidia Corp. also highlighted its Ising family of open models for quantum computing development, with Amazon Web Services integrating Ising into AWS Parallel Computing Service.
ANALYSIS The breadth of the GTC Berlin slate, spanning infrastructure, supply chain, robotics, telecom, enterprise security, and scientific models, positions Nvidia Corp. as an orchestrator across the full AI stack rather than a chip vendor alone.
Nvidia Corp. stock was up 2% in Wednesday premarket trading.