Nvidia: AI is a five-layer cake

Author: Nvidia

Compiled by: PANews

Energy → Chips → Infrastructure → Models → Applications. Every successful application depends on each layer below it, all the way down to the power plants that keep it running.

AI is one of the most powerful forces shaping the world today. It is not a smart application or a single model, but an infrastructure like electricity and the internet.

AI operates on real hardware, real energy, and a real economy. It takes raw materials and transforms them into intelligence on a massive scale. Every company will use it, and every country will build it.

To understand why AI is unfolding in this way, it is helpful to start with first principles and examine the fundamental changes that have occurred in the field of computing.

From pre-recording software to real-time intelligence

For most of computing history, software was pre-recorded. Humans described an algorithm, and computers executed it. Data had to be carefully structured, stored in tables, and retrieved through precise queries. SQL became indispensable because it made the world operational.

AI has disrupted this pattern.

This is the first time we have a computer capable of understanding unstructured information. It can see images, read text, hear sounds, understand meaning, and reason about context and intent. Most importantly, it generates intelligence in real time.

Every response is newly created, and every answer depends on the context you provide. This isn't the software retrieving stored instructions; rather, it's the software reasoning and generating intelligence on demand.

Because intelligence is generated in real time, the entire computing stack beneath it must be reinvented.

AI as infrastructure

From an industrial perspective, AI can be broken down into a five-layer technology stack.

energy

At the very bottom is energy. Real-time generated intelligence requires real-time generated electricity. Every token generated is the result of electronic movement, heat management, and energy conversion into computation. Below this, there is no abstraction layer; energy is the first principle of AI infrastructure and also the upper limit constraining how much intelligence the system can generate.

chip

Above energy are chips. These are processors specifically designed to efficiently convert energy into computation at scale. AI workloads require immense parallelism, high-bandwidth memory, and fast interconnects. Advances at the chip level determine how quickly AI can scale and how affordable intelligence becomes.

Infrastructure

Above the chips lies the infrastructure, including land, power supply, cooling, construction, networks, and the systems that orchestrate tens of thousands of processors into a single machine. These systems are AI factories, not designed to store information, but to create intelligence.

Model

Above the infrastructure are the models. AI models understand a wide variety of information: language, biology, chemistry, physics, finance, medicine, and the physical world itself. Language models are just one category. Some of the most transformative work is taking place in the fields of protein AI, chemical AI, physics simulation, robotics, and autonomous systems.

application

At the very top are the applications, where economic value is created. These include drug discovery platforms, industrial robots, legal assistants, and self-driving cars. Self-driving cars represent AI applications embodied in machines, while humanoid robots represent AI applications embodied in bodies—the same technology stack, different results.

This is the five-layer cake: energy → chip → infrastructure → model → application.

Every successful application depends on every layer below it, all the way down to the power plant that keeps it running.

We've only just begun this construction. Hundreds of billions of dollars have already been invested, but trillions more in infrastructure still need to be built.

Around the world, we are seeing chip factories, computer assembly plants, and AI factories being built on an unprecedented scale. This is becoming the largest infrastructure construction in human history.

The workforce required to support this construction is enormous. AI factories need electricians, plumbers, pipe installers, steelworkers, network technicians, installers, and operators. These are all well-paid, skilled jobs, and in high demand. You don't need a PhD in computer science to participate in this transformation.

At the same time, AI is driving productivity gains in the knowledge economy. Take radiology as an example: AI now assists in interpreting scan results, but the demand for radiologists continues to grow. This is not a paradox.

A radiologist's mission is to care for patients; reading scans is just one task in the process. When AI takes over more routine work, radiologists can focus on judgment, communication, and care. Hospitals become more efficient, serving more patients and hiring more staff. Productivity creates capacity, and capacity creates growth.

What changes have occurred in the past year?

Over the past year, AI has crossed a significant threshold: models have become good enough to provide practical value at scale. Reasoning capabilities have improved, illusions have decreased, and grounding capabilities have significantly increased. For the first time, AI-based applications have begun to generate real economic value.

Applications in drug discovery, logistics, customer service, software development, and manufacturing have demonstrated strong product-market fit, creating a strong pull on each layer below them.

Open-source models play a crucial role here. Most models in the world are free, and researchers, startups, enterprises, and entire nations rely on them for advanced AI. When open-source models reach the forefront, they don't just change the software; they activate the demand across the entire technology stack.

DeepSeek-R1 is a prime example of this. By making a powerful inference model widely available, it accelerates the adoption of application layers and increases the demand for training, infrastructure, chips, and energy beneath them.

what does that mean

The implications become clear when you view AI as infrastructure.

AI began with the Transformer LLM, but it's much more than that. It's an industrial revolution that has reshaped how energy is produced and consumed, how factories are built, how work is organized, and how economies grow.

AI factories are being built because intelligence is now being generated in real time. Chips are being redesigned because efficiency determines how quickly intelligence can scale. Energy becomes central because it sets the ceiling on the total output of intelligence. Applications are accelerating because the models beneath them have finally crossed the threshold of being able to deliver practical value at scale.

Each layer reinforces the others.

This is why this construction is so massive, why it touches so many industries simultaneously, and why it won't be confined to a single country or sector. Every company will use AI, and every country will build it.

We are still in the early stages; most of the infrastructure does not yet exist, most of the workforce is untrained, and most opportunities are yet to be realized.

But the direction is clear.

AI is becoming the infrastructure of the modern world. The choices we make now—how fast we build, how widely we participate, and how responsibly we deploy—will shape the face of this era.