Nvidia unveils AI-factory blueprint and Blackwell-driven systems to industrialise large-scale AI
Introduced Omniverse DSX Blueprint for gigascale data-centre orchestration and throughput optimisation
Revealed NVQLink quantum-GPU interconnect and Nokia collaboration for AI-native telecom infrastructure
Nvidia CEO Jensen Huang at the GTC made a wide-ranging keynote address, positioning the company at the centre of a new wave of AI investment. The chip maker unveiled system and software blueprints for gigascale AI factories, a quantum-GPU interconnect, major telecom partnerships and plans to help build US government AI supercomputers.
Nvidia emphasised that its Blackwell architecture and follow-on systems remain the backbone of its strategy, saying strong demand has created a multiyear backlog tied to tens of millions of GPU sales. The company framed the announcements as a push to improve total cost of ownership for AI workloads by delivering tighter hardware-software co-design across racks and data-centre systems.
Nokia Stake Buy
Nvidia announced a strategic investment and collaboration with Nokia focused on next-generation cellular infrastructure. Nvidia is taking a $1 billion stake in the networking company, the latest partnership for the artificial intelligence chipmaker.
The collaboration aims to focus on AI-native 6G research and telecom software optimised to run on Nvidia silicon. It is part of a broader effort to extend its architecture beyond traditional data centres into telecom and edge networks.
The company said it will work with the US Department of Energy and other partners on a series of national AI supercomputers, underscoring Nvidia’s bid to anchor America’s scientific computing roadmap while expanding its addressable market in government and research.
Blueprint for AI Factories
Nvidia also introduced the Omniverse DSX Blueprint, a packaged set of tools and practices for designing. The company simulates and operates large AI facilities, which the company calls gigawatt-scale “AI factories.”
The blueprint aims to help customers maximise throughput across chips, networking, cooling and software stacks.
A headline technical reveal was NVQLink, an interconnect designed to join quantum processors with Nvidia GPUs for hybrid quantum-GPU workflows and error-correction tasks. Nvidia said dozens of quantum builders and national labs are already on board to test the new interconnect, signalling a push to fold quantum into large-scale scientific computing.
Market Reaction
Investors reacted positively to the product slate and infrastructure narrative. Nvidia shares jumped on the news as markets digested the implications for future GPU demand and new end markets such as telecom, robotics and government-scale deployments.
The company is reportedly attempting to broaden its TAM (total addressable market) by selling not just chips but the designs, software and services for massive AI deployments.
Taken together, the announcements underline Nvidia’s bet that AI will move from isolated models to industrialised, continuously running systems, a shift that favours vendors able to supply integrated hardware, software and operational blueprints.
That positioning raises strategic and regulatory questions about geopolitical supply, US-China access, and how quickly other chipmakers and cloud providers can replicate the integrated stack.
