Jensen Huang unveils five-layer AI architecture "Five-Layer Cake" at GTC 2026 Berlin
At the GTC 2026 Berlin conference, NVIDIA CEO Jensen Huang unveiled a five-layer AI architecture called "Five-Layer Cake," defining AI as a new form of manufacturing that mass-produces tokens. The five layers are, in order: power and energy infrastructure; accelerated computing hardware; AI factory infrastructure (including neoclouds and communication networks); AI models spanning fields such as language, biology, chemistry, physics, robotics, and animation; and the application layer that connects tokens to industries and enterprises. Huang said AI is a factory and machine that autonomously produces intelligence, and called on Europe to build a complete five-layer architecture of its own and shift from being an AI consumer to an AI producer.
Body · AI translation
Nvidia CEO Jensen Huang delivers a keynote address. (Photo: THE ELEC)
PC main text_June Mouser (in Article)//
AD Manager | AD178089411676 //
// AD Manager | AD178089411676
Nvidia has defined artificial intelligence (AI) as a new manufacturing industry that produces digital intelligence, or tokens, at scale. The company also unveiled a five-layer AI architecture, dubbed the "Five-Layer Cake," to support this vision.
Nvidia CEO Jensen Huang said during his keynote at the annual GTC 2026 Berlin conference on Oct. 21 local time, "A new industrial revolution begins when new technology fundamentally changes how products are made and what can be produced." He emphasized that Europe must move beyond being merely a consumer of AI and become a producer of digital intelligence.
Huang described the current period as the beginning of a new industrial revolution and introduced the five-layer AI architecture as a framework for systematically building AI systems.
The first layer consists of power and energy infrastructure that operates AI factories. The second comprises advanced computers and accelerated computing hardware that provide the foundation for AI processing. The third is AI factory infrastructure, combining power and computing to produce tokens at scale, including neoclouds and communication networks.
The fourth layer consists of AI models spanning various domains and scales, including language, biology, chemistry, physics, robotics and animation. The fifth and final layer comprises applications that integrate the resulting tokens into industries and enterprises.
"Unlike conventional software tools, AI is a factory and a machine that produces intelligence on its own," Huang said. He explained that the five-layer AI architecture, connecting energy to applications, would become the core infrastructure of the new industrial revolution. He also urged European companies and governments to build independent production ecosystems rather than remain technology consumers.
"To become a central player in the AI economy, you must produce AI yourself rather than simply consume it," Huang added. "Europe has the capabilities to build the entire five-layer AI architecture independently."