Understand NVIDIA's GTC Conference in One Article: From GPUs to AI Factories, How Jensen Huang is Reinventing American Technological Hegemony

Early in the morning on October 29, 2025, Jensen Huang, the founder and CEO of NVIDIA, delivered a significant speech at the GTC Washington, D.C. Technology Summit held in Washington.

This event, hailed as the "Super Bowl of the AI world," is no longer just a display of technical parameters. Instead, standing in the heart of the U.S. capital, Jensen Huang outlined a grand blueprint for the United States in the "AI Century": from bringing chip manufacturing back to the homeland to leading future communication standards with AI, NVIDIA is trying to help the United States maintain its leadership in AI infrastructure and innovation with the stance of "extreme co - design."

At the beginning of his speech, Jensen Huang pointed directly to the core conflict: in the current situation where Moore's Law has failed and the growth of computing performance has stagnated, NVIDIA is the "savior" that will rescue future computing. They use parallel computing and GPU acceleration to completely liberate the "computing power," the lifeline, from the shackles of traditional CPUs.

01 Paradigm Shift in Computing Architecture: From CPU Dominance to the GPU - Accelerated Era

During his speech, Jensen Huang reviewed the historical turning point in the computing industry. For decades, the performance of CPUs has always followed a predictable trajectory of scale - up growth. However, with the end of Dennard scaling, which holds that by continuously shrinking transistor sizes to maintain power density, power consumption can be reduced and performance can be improved, the traditional development path is no longer sustainable.

In the face of these challenges, NVIDIA's answer is parallel computing, GPUs, and accelerated computing architectures.

Jensen Huang declared, "This turning point has arrived, and NVIDIA is ready. We recognize that by introducing processors that can fully utilize exponentially growing transistors, applying parallel computing technology, and working in tandem with sequentially - processing CPUs, we can push computing power to a whole new dimension - this era has truly come."

The realization of accelerated computing relies on NVIDIA's carefully constructed software foundation - the CUDA - X full - stack acceleration library. This vast software ecosystem covers key areas such as cuDNN and TensorRT - LLM in the field of deep learning, the data science platform RAPIDS (cuDF/cuML), the decision - optimization tool cuOpt, the computational lithography solution cuLitho, and the quantum and hybrid computing frameworks CUDA - Q and cuQuantum.

Jensen Huang hailed this complete software ecosystem as "the company's most precious treasure." It forms the technological core of NVIDIA's accelerated computing strategy, providing the underlying impetus for computing transformation in various industries.

02 AI - Native 6G Technology Stack ARC - Pro: Remodeling the Global Communication Landscape

During his speech, Jensen Huang turned his attention to national security and the economic lifeline. He emphasized that telecommunications technology is the lifeline of the economy and national security, but currently, most global wireless technology deployments rely on the technology systems of other countries.

Jensen Huang said, "This situation where the core communication technology is subject to others must end, and now we have a historical opportunity to turn the situation around." He believes that the United States will "regain the dominance of communication technology."

To achieve this strategic goal, NVIDIA has launched the revolutionary NVIDIA ARC - an AI - native 6G wireless technology stack centered around U.S. technology. This platform is built on NVIDIA's Aerial platform, integrating Grace CPUs, Blackwell GPUs, and advanced network components, and achieving performance breakthroughs through accelerated computing.

According to Jensen Huang, NVIDIA has reached an in - depth strategic cooperation with Nokia, a communication equipment giant. Nokia will integrate the NVIDIA ARC solution into its future base station systems, and this platform will enable key applications such as precise robot control and high - precision weather forecasting.

According to NVIDIA's official blog post, the company plans to integrate its commercial - grade AI - RAN products into Nokia's leading radio access network (RAN) portfolio, enabling communication service providers to deploy AI - native 5G - Advanced and 6G networks on the NVIDIA platform. As part of the cooperation, NVIDIA will subscribe to 166.4 million new Nokia shares at a price of 6.01 euros per share, with a total investment of $1 billion.

03 NVQLink: Unblocking the "Meridians" of Quantum Computing

When discussing the frontiers of computational science, Jensen Huang recalled the vision of quantum physicist Richard Feynman forty years ago: to create a quantum computer that can directly simulate the laws of nature.

"Now we are able to prepare stable, coherent, and error - correcting logical qubits," Jensen Huang pointed out. "However, these qubits are extremely fragile and require powerful technology to support quantum error correction and state interpretation."

To achieve seamless integration between quantum computing and GPU computing, NVIDIA has launched the quantum - GPU interconnection technology NVQLink. This innovation enables the quantum processing unit to call the CUDA - Q computing framework in real - time, reducing communication latency to an extreme level of about 4 microseconds.

During the on - site demonstration, the large screen behind Jensen Huang showed a collaborative innovation network including 17 leading quantum computing companies and multiple U.S. Department of Energy laboratories. "Almost all U.S. Department of Energy - affiliated laboratories are closely collaborating with our quantum computing ecosystem partners to integrate quantum computing into the future scientific development blueprint."

04 Accelerating U.S. Scientific Research: Collaborating with the Department of Energy to Build a New - Generation Supercomputer Cluster

Jensen Huang announced that U.S. national laboratories are entering a new era of scientific research driven by AI infrastructure. NVIDIA has reached a strategic cooperation with the U.S. Department of Energy (DOE) to jointly build seven new - generation supercomputers, providing powerful computing support for future scientific research.

In terms of specific layout, NVIDIA, in collaboration with the U.S. Department of Energy and Oracle, will build the largest AI supercomputer cluster within the DOE system at Argonne National Laboratory.

This supercomputing network consists of two core systems:

The Solstice system will deploy 100,000 NVIDIA Blackwell GPUs. Once completed, it will become the world's largest intelligent agent science platform for public research.

The Equinox system is equipped with 10,000 Blackwell GPUs, providing up to 2,200 EFLOPS of AI computing power, specifically serving cutting - edge scientific computing, simulation, and open research.

This major infrastructure investment marks the United States' official entry into the new era of "intelligent agent - driven scientific research," which will significantly enhance its innovation ability and development speed in key areas such as national security, energy strategy, and basic scientific research.

05 Domestic Manufacturing Strategy: Mass - Producing Blackwell Chips in the United States

In terms of industrial layout, Jensen Huang revealed important progress: the Blackwell GPU has achieved large - scale production in Arizona, USA, and the complete machine systems based on this chip will also be assembled in the United States. This move marks NVIDIA's successful shift from relying entirely on TSMC for manufacturing its flagship products to a domestic supply - chain system in the United States.

It is worth noting that many announcements at this conference have clear policy orientations. By demonstrating its core position in the U.S. technology ecosystem, NVIDIA conveyed a key message to policymakers: restrictions on chip exports will directly harm U.S. interests. Jensen Huang revealed before the conference that Washington was chosen as the conference venue to facilitate President Trump's attendance, but this did not happen due to the president's trip to Asia.

In terms of market performance, Jensen Huang revealed that the demand for GPUs remains strong: NVIDIA has shipped 6 million Blackwell GPUs in the past four quarters, and the total sales of the Blackwell and the next - generation Rubin chips are expected to reach a scale of $500 billion.

06 The AI Factory Revolution: Paradigm Shift from Tools to Productivity Entities

"AI is not a tool but a productivity entity," Jensen Huang put forward this revolutionary view in his speech. "For the first time in history, technology has the ability to perform labor tasks and becomes an extension of human productivity." This fundamental shift from "tools" to "AI workers" is giving rise to a new computing paradigm and, in turn, unprecedented professional forms and industrial landscapes.

In Jensen Huang's vision, the modern "AI factory" is far beyond what a traditional data center can represent. It is a new comprehensive computing platform specifically built for the generation, transmission, and service of massive tokens. This platform - level architecture aims to achieve unprecedented computing density and energy - efficiency ratio.

Facing the exponentially growing demand for AI computing power, Jensen Huang detailed NVIDIA's solution: "First, we redefine the form of the computer, for the first time expanding a single computing system to the scale of an entire cabinet; then, through the innovative AI Ethernet technology Spectrum - X, we achieve lossless horizontal expansion between multiple systems."

With the rise of AI factories, emerging fields such as robotics engineering and quantum science are creating a large number of unprecedented job opportunities. "The innovation flywheel has started," Jensen Huang emphasized. "The next key is to significantly reduce operating costs, optimize the user experience, and maintain the continuous operation of this innovation cycle through cost control."

The key to realizing this vision lies in "extreme co - design," that is, synchronously designing new underlying computing architectures, including chips, system platforms, software stacks, AI models, and terminal applications.

To showcase the physical results of this concept, Jensen Huang presented the new - generation NVIDIA BlueField - 4 DPU on the stage. This data processor, integrating a 64 - core Grace CPU and a ConnectX - 9 network chip, has a computing performance six times that of its predecessor and will become the "operating system core" of future AI factories.

This revolutionary DPU is specifically designed to offload and accelerate network, storage, and security tasks of servers. It is planned to be first deployed on NVIDIA's Vera Rubin rack - level AI platform in 2026 and then opened to a wider server ecosystem.

07 Omniverse DSX: The Ultimate Blueprint for AI Factories

To address the challenges of large - scale AI deployment, Jensen Huang officially launched Omniverse DSX - a comprehensive solution that fully covers the design and operation of AI factories ranging from 100 megawatts to thousands of megawatts. This blueprint has been fully verified at the AI Factory Research Center in Virginia.

To make the DSX reference design more adaptable to different data centers, NVIDIA provides two configuration frameworks:

DSX Boost (Internal Energy - Efficiency Optimization): Through intelligent power management and dynamic workload distribution, it can reduce energy consumption by about 30% with the same computing power output or increase GPU density by 30% with the same power budget, achieving a qualitative leap in token generation throughput.

DSX Flex (External Energy Integration): It deeply integrates the data center into the regional power grid system. By intelligently scheduling renewable energy and balancing supply - and - demand relationships, it can effectively activate about 100 gigawatts of idle capacity in the U.S. power grid.

Omniverse DSX aims to enable new entrants to quickly build AI factories. This solution ensures that the hardware of NVIDIA and its partners is compatible out - of - the - box at the levels of processors, networks, and cooling systems. Even without professional experience, one can deploy according to the blueprint, minimizing customization requirements.

It is worth noting that this architecture not only perfectly supports the current Blackwell platform but also reserves compatibility for future products such as the next - generation Vera Rubin, providing long - term technological guarantees for investors.

08 Open Ecosystem and Industrial Integration: NVIDIA's AI Implementation Strategy

During his speech, Jensen Huang emphasized the core value of an open ecosystem: "Open - source models and open collaboration are the cornerstones of global innovation, providing continuous impetus for startups, research institutions, and industrial companies."

It is reported that NVIDIA has contributed hundreds of high - quality open - source models and datasets to the developer community this year.

NVIDIA has built an open - source model system covering key areas:

Nemotron: Focusing on AI for intelligent agent reasoning and decision - making

Cosmos: Breaking the boundaries of synthetic data generation and physical AI

Isaac GR00T: Enabling robot skill learning and cross - scenario generalization

Clara: Remodeling biomedical research and clinical workflows

These model families will jointly empower the next - generation intelligent agent systems, robotics, and scientific discovery. Jensen Huang emphasized, "We continuously invest in the open ecosystem because it is the common need of scientific research, entrepreneurship, and industrial upgrading."

The achievements of partners demonstrated on - site showed the wide - ranging application scenarios of NVIDIA's technology, covering cloud - computing giants such as Google Cloud, Microsoft Azure, and Oracle, enterprise service providers such as ServiceNow and SAP, and leaders in professional fields such as Synopsys and Cadence.

Jensen Huang also announced two strategic cooperations:

Collaborating with CrowdStrike to build a new - generation network security system, achieving "light - speed" threat detection and response from the cloud to the edge through the Nemotron model and the NeMo toolchain;

Reaching an in - depth technical integration with Palantir, integrating the accelerated computing architecture, CUDA - X library, and open - source models into the Ontology data platform to achieve performance breakthroughs in large - scale data processing.

09 Physical AI: Digital Twin - Driven Industrial Transformation

"Physical AI" is driving the process of re - industrialization in the United States, that is, reshaping manufacturing, logistics, and infrastructure through robots and intelligent