Jensen Huang, CEO of NVIDIA, proclaims a new era at COMPUTEX 2025: Artificial Intelligence breaks the classic rules of silicon and accelerates its development at an unprecedented pace.
With a statement as provocative as it is visionary, Jensen Huang made it clear that the model guiding semiconductor evolution for over half a century no longer dictates the pace of innovation. “Moore’s Law is over. In the future, only the sky can limit the speed of development in the AI industry,” the executive declared before an audience filled with tech leaders and developers at the event’s conclusion.
Joined by Liu Yangwei, president of Foxconn, Huang presented his vision for the future of computing and technology infrastructure, marking a significant turning point. While chip manufacturers still struggle to advance in increasingly complex and costly manufacturing nodes, NVIDIA has chosen an alternative path: integrating the entire tech ecosystem— from hardware to AI models— to scale without relying on transistor size.
A New Architecture: From Silicon to Collaborative Superchip
Beyond creating smaller chips, NVIDIA is betting on technologies like 3D packaging and NVLink, its ultra-high-speed interconnect that turns multiple components into a single collaborative “superchip.” This approach not only overcomes classic bottlenecks in computing but also establishes a new paradigm of systemic architecture, where GPUs, networks, and storage operate as a single organism.
According to Huang, the traditional approach is no longer sufficient. “It’s no longer about designing a faster chip, but about redesigning the entire system—from models to data centers—to adapt to the new engine of change: artificial intelligence.”
Goodbye to Moore’s Law, Welcome to Huang’s Law
Moore’s Law, formulated in the 1960s, posited that the number of transistors on a chip would double approximately every two years, increasing performance at lower cost. However, physical limits and the energy costs of current processes—such as the 3 nm and future 2 nm nodes—have slowed this progress.
In response, Huang proposes a different vision, already known in the industry as “Huang’s Law,” based not on lithographic scaling but rather on exponential performance acceleration through vertical integration, AI-optimized software, and collaboration among chips. And rightly so: NVIDIA’s growth has not slowed. The company has transformed into the backbone of the new AI economy, with revenues positioning it as one of the most influential players of the 21st century.
The New Ecosystem: Redesigning from Scratch
During his speech, Huang emphasized that the real challenge lies not just in chip architecture but in the complete redesign of the tech ecosystem:
- New GPUs designed for generative AI, not just for graphics.
- Data centers designed for foundation models with trillions of parameters.
- Photonic networks and quantum computing on the horizon.
- Software that dynamically adapts to training and deployment models in real-time.
And in that future, Huang hinted at an even more disruptive possibility: AI designing future GPUs. The role of the human engineer will shift towards overseeing infrastructures and designing neural architectures, with automation and AI handling the more technical stages of electronic design.
A Warning to the Industry: Adapt or Disappear
NVIDIA’s message was not just technical; it was strategic. “Anyone wishing to keep pace with AI will have to abandon the old rules of the game,” Huang warned. Competing in manufacturing nodes is no longer enough. The real differentiator lies in the ability to adapt to disruption, vertically integrate technology and vision, and scale beyond hardware.
In the post-Moore era, where energy costs and chip size are no longer the only relevant metrics, the real limit—as NVIDIA’s CEO stated—may not lie in physics, but rather in our own ability to imagine how far artificial intelligence can go.