The author analyzes Jensen Huang's blog post on AI's five-layer architecture and presents it as NVIDIA's expansion blueprint, expressing bullish views on AI infrastructure growth and NVIDIA's ecosystem positioning.

Jensen Huang published a blog post in his personal capacity after many years, titled "AI is a five layer cake". Jensen Huang is a man of few words who rarely updates his blog; previous updates mostly discussed his views on NVIDIA's acquisitions. I think this content can be compared to the post "The Intelligent Industrial Revolution" updated in October 2016. If 2016 was NVIDIA's transformation manifesto, then this post is NVIDIA's expansion blueprint and logical closed loop. Below is a breakdown of the content of this blog post: 1. AI's Five-Layer Architecture: Energy → Chips → Infrastructure → Models → Applications, with Energy at the bottom layer. AI infrastructure is still in its early stages; hundreds of billions of dollars have been invested so far, but infrastructure worth trillions of dollars still needs to be built; This industrial revolution also requires high-end software engineers, as well as a vast number of electricians, plumbers, steelworkers, and network technicians to participate in the construction of AI factories. Power infrastructure, such as power grids, nuclear energy, energy storage, and green energy, will become the most critical external support in NVIDIA's ecosystem. 2. AI is the Ultimate Productivity Force: AI is not meant to replace humans, but to expand the total market size by reducing costs and increasing efficiency. Taking radiology as an example, AI is already able to assist in interpreting scan images, but the demand for radiologists continues to grow. When AI takes on more routine tasks, radiologists can focus on judgment, communication, and care. Hospital efficiency will increase, enabling them to serve more patients; 3. Strong Pull Effect of Applications: In the past year, AI has crossed an important threshold, and applications built on AI have started creating real economic value for the first time. Applications in drug development, logistics, customer service, software development, and manufacturing have already demonstrated strong product-market fit. When every application feeds back and pulls every layer of the architecture, the industry will generate an endogenous high-speed growth inertia; 4. Mutual Reinforcement Between Layers: Every successful application at the upper layer will pull all links at the lower layers downward, extending all the way to the power plants at the bottom layer. Jensen Huang specifically affirmed the huge contribution of open-source models in the article. Taking DeepSeek-R1 as an example, he pointed out that when this powerful reasoning model was widely open-sourced, it greatly accelerated the popularization of top-layer applications, thereby directly triggering huge demand at the bottom layer for training compute, infrastructure, chips, and energy. 5. Sovereign AI and Agents: Software forms are transforming from passively executed functional tools to proactively closed-loop AI Agents. In the future, AI will reshape the production methods of energy, the construction models of factories, and the growth paths of the economy. Therefore, every company will use AI, and every country will build its own AI infrastructure.